| { | |
| "model.lm_head.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0000_Of_20.pt", | |
| "model.model.embed_tokens.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0001_Of_20.pt", | |
| "model.model.layers.0.mlp.down_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0002_Of_20.pt", | |
| "model.model.layers.14.mlp.down_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0002_Of_20.pt", | |
| "model.model.layers.2.mlp.down_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0002_Of_20.pt", | |
| "model.model.layers.25.mlp.down_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0002_Of_20.pt", | |
| "model.model.layers.6.mlp.down_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0002_Of_20.pt", | |
| "model.model.layers.17.self_attn.time_mixer.key.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0002_Of_20.pt", | |
| "model.model.layers.20.self_attn.time_mixer.receptance.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0002_Of_20.pt", | |
| "model.model.layers.25.self_attn.time_mixer.key.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0002_Of_20.pt", | |
| "model.model.layers.4.self_attn.time_mixer.receptance.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0002_Of_20.pt", | |
| "model.model.layers.9.self_attn.time_mixer.key.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0002_Of_20.pt", | |
| "model.model.layers.0.mlp.gate_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0003_Of_20.pt", | |
| "model.model.layers.14.mlp.gate_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0003_Of_20.pt", | |
| "model.model.layers.2.mlp.gate_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0003_Of_20.pt", | |
| "model.model.layers.25.mlp.gate_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0003_Of_20.pt", | |
| "model.model.layers.6.mlp.gate_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0003_Of_20.pt", | |
| "model.model.layers.17.self_attn.time_mixer.output.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0003_Of_20.pt", | |
| "model.model.layers.20.self_attn.time_mixer.value.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0003_Of_20.pt", | |
| "model.model.layers.25.self_attn.time_mixer.output.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0003_Of_20.pt", | |
| "model.model.layers.4.self_attn.time_mixer.value.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0003_Of_20.pt", | |
| "model.model.layers.9.self_attn.time_mixer.output.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0003_Of_20.pt", | |
| "model.model.layers.0.mlp.up_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0004_Of_20.pt", | |
| "model.model.layers.14.mlp.up_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0004_Of_20.pt", | |
| "model.model.layers.2.mlp.up_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0004_Of_20.pt", | |
| "model.model.layers.25.mlp.up_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0004_Of_20.pt", | |
| "model.model.layers.6.mlp.up_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0004_Of_20.pt", | |
| "model.model.layers.17.self_attn.time_mixer.receptance.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0004_Of_20.pt", | |
| "model.model.layers.21.self_attn.time_mixer.key.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0004_Of_20.pt", | |
| "model.model.layers.25.self_attn.time_mixer.receptance.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0004_Of_20.pt", | |
| "model.model.layers.5.self_attn.time_mixer.key.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0004_Of_20.pt", | |
| "model.model.layers.9.self_attn.time_mixer.receptance.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0004_Of_20.pt", | |
| "model.model.layers.1.mlp.down_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0005_Of_20.pt", | |
| "model.model.layers.15.mlp.down_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0005_Of_20.pt", | |
| "model.model.layers.20.mlp.down_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0005_Of_20.pt", | |
| "model.model.layers.26.mlp.down_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0005_Of_20.pt", | |
| "model.model.layers.7.mlp.down_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0005_Of_20.pt", | |
| "model.model.layers.17.self_attn.time_mixer.value.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0005_Of_20.pt", | |
| "model.model.layers.21.self_attn.time_mixer.output.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0005_Of_20.pt", | |
| "model.model.layers.25.self_attn.time_mixer.value.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0005_Of_20.pt", | |
| "model.model.layers.5.self_attn.time_mixer.output.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0005_Of_20.pt", | |
| "model.model.layers.9.self_attn.time_mixer.value.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0005_Of_20.pt", | |
| "model.model.layers.1.mlp.gate_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.15.mlp.gate_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.20.mlp.gate_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.26.mlp.gate_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.7.mlp.gate_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.18.self_attn.time_mixer.key.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.21.self_attn.time_mixer.receptance.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.26.self_attn.time_mixer.key.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.5.self_attn.time_mixer.receptance.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.0.self_attn.time_mixer.g1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.12.self_attn.time_mixer.g1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.16.self_attn.time_mixer.g1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.2.self_attn.time_mixer.g1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.23.self_attn.time_mixer.g1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.27.self_attn.time_mixer.g1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.6.self_attn.time_mixer.g1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.0.self_attn.time_mixer.a1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.10.self_attn.time_mixer.a1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.12.self_attn.time_mixer.a1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.14.self_attn.time_mixer.a1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.16.self_attn.time_mixer.a1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.18.self_attn.time_mixer.a1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.2.self_attn.time_mixer.a1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.21.self_attn.time_mixer.a1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.23.self_attn.time_mixer.a1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.25.self_attn.time_mixer.a1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.27.self_attn.time_mixer.a1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.4.self_attn.time_mixer.a1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.6.self_attn.time_mixer.a1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.8.self_attn.time_mixer.a1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.0.self_attn.time_mixer.v1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.12.self_attn.time_mixer.v1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.16.self_attn.time_mixer.v1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.2.self_attn.time_mixer.v1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.23.self_attn.time_mixer.v1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.27.self_attn.time_mixer.v1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.6.self_attn.time_mixer.v1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.0.input_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.0.self_attn.time_mixer.w0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.1.post_attention_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.1.self_attn.time_mixer.x_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.10.self_attn.time_mixer.a0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.10.self_attn.time_mixer.x_g": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.11.self_attn.time_mixer.k_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.11.self_attn.time_mixer.x_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.12.self_attn.time_mixer.k_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.12.self_attn.time_mixer.x_r": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.13.self_attn.time_mixer.ln_x.bias": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.13.self_attn.time_mixer.x_v": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.14.self_attn.time_mixer.ln_x.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.14.self_attn.time_mixer.x_w": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.15.self_attn.time_mixer.v0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.16.input_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.16.self_attn.time_mixer.w0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.17.post_attention_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.17.self_attn.time_mixer.x_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.18.self_attn.time_mixer.a0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.18.self_attn.time_mixer.x_g": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.19.self_attn.time_mixer.k_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.19.self_attn.time_mixer.x_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.2.self_attn.time_mixer.k_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.2.self_attn.time_mixer.x_r": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.20.self_attn.time_mixer.ln_x.bias": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.20.self_attn.time_mixer.x_v": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.21.self_attn.time_mixer.ln_x.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.21.self_attn.time_mixer.x_w": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.22.self_attn.time_mixer.v0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.23.input_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.23.self_attn.time_mixer.w0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.24.post_attention_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.24.self_attn.time_mixer.x_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.25.self_attn.time_mixer.a0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.25.self_attn.time_mixer.x_g": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.26.self_attn.time_mixer.k_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.26.self_attn.time_mixer.x_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.27.self_attn.time_mixer.k_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.27.self_attn.time_mixer.x_r": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.3.self_attn.time_mixer.ln_x.bias": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.3.self_attn.time_mixer.x_v": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.4.self_attn.time_mixer.ln_x.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.4.self_attn.time_mixer.x_w": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.5.self_attn.time_mixer.v0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.6.input_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.6.self_attn.time_mixer.w0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.7.post_attention_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.7.self_attn.time_mixer.x_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.8.self_attn.time_mixer.a0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.8.self_attn.time_mixer.x_g": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.9.self_attn.time_mixer.k_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.9.self_attn.time_mixer.x_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.11.self_attn.time_mixer.r_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.19.self_attn.time_mixer.r_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.26.self_attn.time_mixer.r_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.9.self_attn.time_mixer.r_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0006_Of_20.pt", | |
| "model.model.layers.1.mlp.up_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.15.mlp.up_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.20.mlp.up_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.26.mlp.up_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.7.mlp.up_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.18.self_attn.time_mixer.output.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.21.self_attn.time_mixer.value.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.26.self_attn.time_mixer.output.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.5.self_attn.time_mixer.value.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.0.self_attn.time_mixer.g2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.12.self_attn.time_mixer.g2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.16.self_attn.time_mixer.g2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.2.self_attn.time_mixer.g2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.23.self_attn.time_mixer.g2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.27.self_attn.time_mixer.g2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.6.self_attn.time_mixer.g2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.0.self_attn.time_mixer.a2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.10.self_attn.time_mixer.a2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.12.self_attn.time_mixer.a2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.14.self_attn.time_mixer.a2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.16.self_attn.time_mixer.a2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.18.self_attn.time_mixer.a2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.2.self_attn.time_mixer.a2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.21.self_attn.time_mixer.a2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.23.self_attn.time_mixer.a2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.25.self_attn.time_mixer.a2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.27.self_attn.time_mixer.a2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.4.self_attn.time_mixer.a2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.6.self_attn.time_mixer.a2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.8.self_attn.time_mixer.a2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.0.self_attn.time_mixer.v2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.12.self_attn.time_mixer.v2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.16.self_attn.time_mixer.v2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.2.self_attn.time_mixer.v2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.23.self_attn.time_mixer.v2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.27.self_attn.time_mixer.v2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.6.self_attn.time_mixer.v2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.0.post_attention_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.0.self_attn.time_mixer.x_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.1.self_attn.time_mixer.a0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.1.self_attn.time_mixer.x_g": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.10.self_attn.time_mixer.k_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.10.self_attn.time_mixer.x_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.11.self_attn.time_mixer.k_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.11.self_attn.time_mixer.x_r": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.12.self_attn.time_mixer.ln_x.bias": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.12.self_attn.time_mixer.x_v": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.13.self_attn.time_mixer.ln_x.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.13.self_attn.time_mixer.x_w": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.14.self_attn.time_mixer.v0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.15.input_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.15.self_attn.time_mixer.w0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.16.post_attention_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.16.self_attn.time_mixer.x_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.17.self_attn.time_mixer.a0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.17.self_attn.time_mixer.x_g": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.18.self_attn.time_mixer.k_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.18.self_attn.time_mixer.x_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.19.self_attn.time_mixer.k_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.19.self_attn.time_mixer.x_r": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.2.self_attn.time_mixer.ln_x.bias": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.2.self_attn.time_mixer.x_v": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.20.self_attn.time_mixer.ln_x.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.20.self_attn.time_mixer.x_w": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.21.self_attn.time_mixer.v0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.22.input_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.22.self_attn.time_mixer.w0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.23.post_attention_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.23.self_attn.time_mixer.x_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.24.self_attn.time_mixer.a0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.24.self_attn.time_mixer.x_g": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.25.self_attn.time_mixer.k_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.25.self_attn.time_mixer.x_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.26.self_attn.time_mixer.k_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.26.self_attn.time_mixer.x_r": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.27.self_attn.time_mixer.ln_x.bias": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.27.self_attn.time_mixer.x_v": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.3.self_attn.time_mixer.ln_x.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.3.self_attn.time_mixer.x_w": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.4.self_attn.time_mixer.v0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.5.input_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.5.self_attn.time_mixer.w0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.6.post_attention_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.6.self_attn.time_mixer.x_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.7.self_attn.time_mixer.a0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.7.self_attn.time_mixer.x_g": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.8.self_attn.time_mixer.k_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.8.self_attn.time_mixer.x_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.9.self_attn.time_mixer.k_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.9.self_attn.time_mixer.x_r": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.12.self_attn.time_mixer.r_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.2.self_attn.time_mixer.r_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.27.self_attn.time_mixer.r_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0007_Of_20.pt", | |
| "model.model.layers.10.mlp.down_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.16.mlp.down_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.21.mlp.down_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.27.mlp.down_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.8.mlp.down_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.18.self_attn.time_mixer.receptance.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.22.self_attn.time_mixer.key.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.26.self_attn.time_mixer.receptance.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.6.self_attn.time_mixer.key.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.1.self_attn.time_mixer.g1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.13.self_attn.time_mixer.g1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.17.self_attn.time_mixer.g1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.20.self_attn.time_mixer.g1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.24.self_attn.time_mixer.g1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.3.self_attn.time_mixer.g1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.7.self_attn.time_mixer.g1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.0.self_attn.time_mixer.w1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.10.self_attn.time_mixer.w1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.12.self_attn.time_mixer.w1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.14.self_attn.time_mixer.w1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.16.self_attn.time_mixer.w1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.18.self_attn.time_mixer.w1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.2.self_attn.time_mixer.w1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.21.self_attn.time_mixer.w1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.23.self_attn.time_mixer.w1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.25.self_attn.time_mixer.w1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.27.self_attn.time_mixer.w1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.4.self_attn.time_mixer.w1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.6.self_attn.time_mixer.w1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.8.self_attn.time_mixer.w1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.1.self_attn.time_mixer.v1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.13.self_attn.time_mixer.v1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.17.self_attn.time_mixer.v1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.20.self_attn.time_mixer.v1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.24.self_attn.time_mixer.v1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.3.self_attn.time_mixer.v1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.7.self_attn.time_mixer.v1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.0.self_attn.time_mixer.a0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.0.self_attn.time_mixer.x_g": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.1.self_attn.time_mixer.k_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.1.self_attn.time_mixer.x_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.10.self_attn.time_mixer.k_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.10.self_attn.time_mixer.x_r": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.11.self_attn.time_mixer.ln_x.bias": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.11.self_attn.time_mixer.x_v": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.12.self_attn.time_mixer.ln_x.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.12.self_attn.time_mixer.x_w": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.13.self_attn.time_mixer.v0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.14.input_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.14.self_attn.time_mixer.w0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.15.post_attention_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.15.self_attn.time_mixer.x_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.16.self_attn.time_mixer.a0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.16.self_attn.time_mixer.x_g": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.17.self_attn.time_mixer.k_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.17.self_attn.time_mixer.x_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.18.self_attn.time_mixer.k_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.18.self_attn.time_mixer.x_r": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.19.self_attn.time_mixer.ln_x.bias": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.19.self_attn.time_mixer.x_v": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.2.self_attn.time_mixer.ln_x.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.2.self_attn.time_mixer.x_w": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.20.self_attn.time_mixer.v0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.21.input_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.21.self_attn.time_mixer.w0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.22.post_attention_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.22.self_attn.time_mixer.x_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.23.self_attn.time_mixer.a0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.23.self_attn.time_mixer.x_g": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.24.self_attn.time_mixer.k_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.24.self_attn.time_mixer.x_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.25.self_attn.time_mixer.k_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.25.self_attn.time_mixer.x_r": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.26.self_attn.time_mixer.ln_x.bias": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.26.self_attn.time_mixer.x_v": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.27.self_attn.time_mixer.ln_x.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.27.self_attn.time_mixer.x_w": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.3.self_attn.time_mixer.v0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.4.input_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.4.self_attn.time_mixer.w0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.5.post_attention_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.5.self_attn.time_mixer.x_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.6.self_attn.time_mixer.a0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.6.self_attn.time_mixer.x_g": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.7.self_attn.time_mixer.k_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.7.self_attn.time_mixer.x_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.8.self_attn.time_mixer.k_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.8.self_attn.time_mixer.x_r": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.9.self_attn.time_mixer.ln_x.bias": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.9.self_attn.time_mixer.x_v": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.13.self_attn.time_mixer.r_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.20.self_attn.time_mixer.r_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.3.self_attn.time_mixer.r_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0008_Of_20.pt", | |
| "model.model.layers.10.mlp.gate_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.16.mlp.gate_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.21.mlp.gate_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.27.mlp.gate_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.8.mlp.gate_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.18.self_attn.time_mixer.value.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.22.self_attn.time_mixer.output.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.26.self_attn.time_mixer.value.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.6.self_attn.time_mixer.output.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.1.self_attn.time_mixer.g2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.13.self_attn.time_mixer.g2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.17.self_attn.time_mixer.g2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.20.self_attn.time_mixer.g2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.24.self_attn.time_mixer.g2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.3.self_attn.time_mixer.g2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.7.self_attn.time_mixer.g2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.0.self_attn.time_mixer.w2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.10.self_attn.time_mixer.w2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.12.self_attn.time_mixer.w2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.14.self_attn.time_mixer.w2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.16.self_attn.time_mixer.w2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.18.self_attn.time_mixer.w2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.2.self_attn.time_mixer.w2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.21.self_attn.time_mixer.w2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.23.self_attn.time_mixer.w2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.25.self_attn.time_mixer.w2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.27.self_attn.time_mixer.w2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.4.self_attn.time_mixer.w2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.6.self_attn.time_mixer.w2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.8.self_attn.time_mixer.w2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.1.self_attn.time_mixer.v2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.13.self_attn.time_mixer.v2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.17.self_attn.time_mixer.v2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.20.self_attn.time_mixer.v2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.24.self_attn.time_mixer.v2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.3.self_attn.time_mixer.v2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.7.self_attn.time_mixer.v2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.0.self_attn.time_mixer.k_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.0.self_attn.time_mixer.x_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.1.self_attn.time_mixer.k_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.1.self_attn.time_mixer.x_r": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.10.self_attn.time_mixer.ln_x.bias": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.10.self_attn.time_mixer.x_v": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.11.self_attn.time_mixer.ln_x.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.11.self_attn.time_mixer.x_w": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.12.self_attn.time_mixer.v0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.13.input_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.13.self_attn.time_mixer.w0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.14.post_attention_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.14.self_attn.time_mixer.x_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.15.self_attn.time_mixer.a0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.15.self_attn.time_mixer.x_g": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.16.self_attn.time_mixer.k_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.16.self_attn.time_mixer.x_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.17.self_attn.time_mixer.k_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.17.self_attn.time_mixer.x_r": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.18.self_attn.time_mixer.ln_x.bias": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.18.self_attn.time_mixer.x_v": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.19.self_attn.time_mixer.ln_x.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.19.self_attn.time_mixer.x_w": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.2.self_attn.time_mixer.v0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.20.input_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.20.self_attn.time_mixer.w0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.21.post_attention_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.21.self_attn.time_mixer.x_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.22.self_attn.time_mixer.a0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.22.self_attn.time_mixer.x_g": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.23.self_attn.time_mixer.k_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.23.self_attn.time_mixer.x_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.24.self_attn.time_mixer.k_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.24.self_attn.time_mixer.x_r": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.25.self_attn.time_mixer.ln_x.bias": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.25.self_attn.time_mixer.x_v": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.26.self_attn.time_mixer.ln_x.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.26.self_attn.time_mixer.x_w": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.27.self_attn.time_mixer.v0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.3.input_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.3.self_attn.time_mixer.w0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.4.post_attention_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.4.self_attn.time_mixer.x_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.5.self_attn.time_mixer.a0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.5.self_attn.time_mixer.x_g": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.6.self_attn.time_mixer.k_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.6.self_attn.time_mixer.x_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.7.self_attn.time_mixer.k_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.7.self_attn.time_mixer.x_r": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.8.self_attn.time_mixer.ln_x.bias": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.8.self_attn.time_mixer.x_v": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.9.self_attn.time_mixer.ln_x.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.9.self_attn.time_mixer.x_w": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.14.self_attn.time_mixer.r_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.21.self_attn.time_mixer.r_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.4.self_attn.time_mixer.r_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0009_Of_20.pt", | |
| "model.model.layers.10.mlp.up_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.16.mlp.up_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.21.mlp.up_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.27.mlp.up_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.8.mlp.up_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.19.self_attn.time_mixer.key.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.22.self_attn.time_mixer.receptance.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.27.self_attn.time_mixer.key.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.6.self_attn.time_mixer.receptance.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.10.self_attn.time_mixer.g1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.14.self_attn.time_mixer.g1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.18.self_attn.time_mixer.g1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.21.self_attn.time_mixer.g1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.25.self_attn.time_mixer.g1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.4.self_attn.time_mixer.g1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.8.self_attn.time_mixer.g1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.1.self_attn.time_mixer.a1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.11.self_attn.time_mixer.a1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.13.self_attn.time_mixer.a1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.15.self_attn.time_mixer.a1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.17.self_attn.time_mixer.a1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.19.self_attn.time_mixer.a1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.20.self_attn.time_mixer.a1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.22.self_attn.time_mixer.a1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.24.self_attn.time_mixer.a1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.26.self_attn.time_mixer.a1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.3.self_attn.time_mixer.a1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.5.self_attn.time_mixer.a1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.7.self_attn.time_mixer.a1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.9.self_attn.time_mixer.a1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.10.self_attn.time_mixer.v1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.14.self_attn.time_mixer.v1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.18.self_attn.time_mixer.v1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.21.self_attn.time_mixer.v1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.25.self_attn.time_mixer.v1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.4.self_attn.time_mixer.v1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.8.self_attn.time_mixer.v1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.0.self_attn.time_mixer.k_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.0.self_attn.time_mixer.x_r": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.1.self_attn.time_mixer.ln_x.bias": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.1.self_attn.time_mixer.x_v": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.10.self_attn.time_mixer.ln_x.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.10.self_attn.time_mixer.x_w": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.11.self_attn.time_mixer.v0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.12.input_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.12.self_attn.time_mixer.w0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.13.post_attention_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.13.self_attn.time_mixer.x_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.14.self_attn.time_mixer.a0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.14.self_attn.time_mixer.x_g": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.15.self_attn.time_mixer.k_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.15.self_attn.time_mixer.x_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.16.self_attn.time_mixer.k_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.16.self_attn.time_mixer.x_r": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.17.self_attn.time_mixer.ln_x.bias": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.17.self_attn.time_mixer.x_v": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.18.self_attn.time_mixer.ln_x.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.18.self_attn.time_mixer.x_w": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.19.self_attn.time_mixer.v0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.2.input_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.2.self_attn.time_mixer.w0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.20.post_attention_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.20.self_attn.time_mixer.x_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.21.self_attn.time_mixer.a0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.21.self_attn.time_mixer.x_g": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.22.self_attn.time_mixer.k_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.22.self_attn.time_mixer.x_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.23.self_attn.time_mixer.k_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.23.self_attn.time_mixer.x_r": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.24.self_attn.time_mixer.ln_x.bias": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.24.self_attn.time_mixer.x_v": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.25.self_attn.time_mixer.ln_x.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.25.self_attn.time_mixer.x_w": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.26.self_attn.time_mixer.v0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.27.input_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.27.self_attn.time_mixer.w0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.3.post_attention_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.3.self_attn.time_mixer.x_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.4.self_attn.time_mixer.a0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.4.self_attn.time_mixer.x_g": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.5.self_attn.time_mixer.k_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.5.self_attn.time_mixer.x_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.6.self_attn.time_mixer.k_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.6.self_attn.time_mixer.x_r": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.7.self_attn.time_mixer.ln_x.bias": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.7.self_attn.time_mixer.x_v": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.8.self_attn.time_mixer.ln_x.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.8.self_attn.time_mixer.x_w": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.9.self_attn.time_mixer.v0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.norm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.15.self_attn.time_mixer.r_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.22.self_attn.time_mixer.r_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.5.self_attn.time_mixer.r_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0010_Of_20.pt", | |
| "model.model.layers.11.mlp.down_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.17.mlp.down_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.22.mlp.down_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.3.mlp.down_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.9.mlp.down_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.19.self_attn.time_mixer.output.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.22.self_attn.time_mixer.value.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.27.self_attn.time_mixer.output.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.6.self_attn.time_mixer.value.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.10.self_attn.time_mixer.g2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.14.self_attn.time_mixer.g2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.18.self_attn.time_mixer.g2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.21.self_attn.time_mixer.g2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.25.self_attn.time_mixer.g2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.4.self_attn.time_mixer.g2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.8.self_attn.time_mixer.g2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.1.self_attn.time_mixer.a2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.11.self_attn.time_mixer.a2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.13.self_attn.time_mixer.a2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.15.self_attn.time_mixer.a2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.17.self_attn.time_mixer.a2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.19.self_attn.time_mixer.a2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.20.self_attn.time_mixer.a2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.22.self_attn.time_mixer.a2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.24.self_attn.time_mixer.a2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.26.self_attn.time_mixer.a2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.3.self_attn.time_mixer.a2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.5.self_attn.time_mixer.a2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.7.self_attn.time_mixer.a2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.9.self_attn.time_mixer.a2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.10.self_attn.time_mixer.v2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.14.self_attn.time_mixer.v2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.18.self_attn.time_mixer.v2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.21.self_attn.time_mixer.v2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.25.self_attn.time_mixer.v2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.4.self_attn.time_mixer.v2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.8.self_attn.time_mixer.v2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.0.self_attn.time_mixer.ln_x.bias": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.0.self_attn.time_mixer.x_v": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.1.self_attn.time_mixer.ln_x.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.1.self_attn.time_mixer.x_w": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.10.self_attn.time_mixer.v0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.11.input_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.11.self_attn.time_mixer.w0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.12.post_attention_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.12.self_attn.time_mixer.x_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.13.self_attn.time_mixer.a0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.13.self_attn.time_mixer.x_g": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.14.self_attn.time_mixer.k_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.14.self_attn.time_mixer.x_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.15.self_attn.time_mixer.k_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.15.self_attn.time_mixer.x_r": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.16.self_attn.time_mixer.ln_x.bias": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.16.self_attn.time_mixer.x_v": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.17.self_attn.time_mixer.ln_x.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.17.self_attn.time_mixer.x_w": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.18.self_attn.time_mixer.v0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.19.input_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.19.self_attn.time_mixer.w0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.2.post_attention_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.2.self_attn.time_mixer.x_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.20.self_attn.time_mixer.a0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.20.self_attn.time_mixer.x_g": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.21.self_attn.time_mixer.k_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.21.self_attn.time_mixer.x_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.22.self_attn.time_mixer.k_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.22.self_attn.time_mixer.x_r": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.23.self_attn.time_mixer.ln_x.bias": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.23.self_attn.time_mixer.x_v": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.24.self_attn.time_mixer.ln_x.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.24.self_attn.time_mixer.x_w": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.25.self_attn.time_mixer.v0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.26.input_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.26.self_attn.time_mixer.w0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.27.post_attention_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.27.self_attn.time_mixer.x_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.3.self_attn.time_mixer.a0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.3.self_attn.time_mixer.x_g": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.4.self_attn.time_mixer.k_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.4.self_attn.time_mixer.x_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.5.self_attn.time_mixer.k_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.5.self_attn.time_mixer.x_r": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.6.self_attn.time_mixer.ln_x.bias": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.6.self_attn.time_mixer.x_v": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.7.self_attn.time_mixer.ln_x.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.7.self_attn.time_mixer.x_w": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.8.self_attn.time_mixer.v0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.9.input_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.9.self_attn.time_mixer.w0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.0.self_attn.time_mixer.r_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.16.self_attn.time_mixer.r_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.23.self_attn.time_mixer.r_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.6.self_attn.time_mixer.r_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0011_Of_20.pt", | |
| "model.model.layers.11.mlp.gate_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.17.mlp.gate_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.22.mlp.gate_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.3.mlp.gate_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.9.mlp.gate_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.19.self_attn.time_mixer.receptance.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.23.self_attn.time_mixer.key.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.27.self_attn.time_mixer.receptance.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.7.self_attn.time_mixer.key.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.11.self_attn.time_mixer.g1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.15.self_attn.time_mixer.g1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.19.self_attn.time_mixer.g1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.22.self_attn.time_mixer.g1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.26.self_attn.time_mixer.g1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.5.self_attn.time_mixer.g1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.9.self_attn.time_mixer.g1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.1.self_attn.time_mixer.w1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.11.self_attn.time_mixer.w1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.13.self_attn.time_mixer.w1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.15.self_attn.time_mixer.w1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.17.self_attn.time_mixer.w1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.19.self_attn.time_mixer.w1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.20.self_attn.time_mixer.w1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.22.self_attn.time_mixer.w1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.24.self_attn.time_mixer.w1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.26.self_attn.time_mixer.w1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.3.self_attn.time_mixer.w1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.5.self_attn.time_mixer.w1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.7.self_attn.time_mixer.w1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.9.self_attn.time_mixer.w1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.11.self_attn.time_mixer.v1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.15.self_attn.time_mixer.v1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.19.self_attn.time_mixer.v1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.22.self_attn.time_mixer.v1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.26.self_attn.time_mixer.v1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.5.self_attn.time_mixer.v1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.9.self_attn.time_mixer.v1": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.0.self_attn.time_mixer.ln_x.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.0.self_attn.time_mixer.x_w": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.1.self_attn.time_mixer.v0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.10.input_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.10.self_attn.time_mixer.w0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.11.post_attention_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.11.self_attn.time_mixer.x_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.12.self_attn.time_mixer.a0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.12.self_attn.time_mixer.x_g": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.13.self_attn.time_mixer.k_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.13.self_attn.time_mixer.x_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.14.self_attn.time_mixer.k_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.14.self_attn.time_mixer.x_r": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.15.self_attn.time_mixer.ln_x.bias": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.15.self_attn.time_mixer.x_v": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.16.self_attn.time_mixer.ln_x.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.16.self_attn.time_mixer.x_w": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.17.self_attn.time_mixer.v0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.18.input_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.18.self_attn.time_mixer.w0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.19.post_attention_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.19.self_attn.time_mixer.x_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.2.self_attn.time_mixer.a0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.2.self_attn.time_mixer.x_g": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.20.self_attn.time_mixer.k_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.20.self_attn.time_mixer.x_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.21.self_attn.time_mixer.k_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.21.self_attn.time_mixer.x_r": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.22.self_attn.time_mixer.ln_x.bias": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.22.self_attn.time_mixer.x_v": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.23.self_attn.time_mixer.ln_x.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.23.self_attn.time_mixer.x_w": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.24.self_attn.time_mixer.v0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.25.input_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.25.self_attn.time_mixer.w0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.26.post_attention_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.26.self_attn.time_mixer.x_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.27.self_attn.time_mixer.a0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.27.self_attn.time_mixer.x_g": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.3.self_attn.time_mixer.k_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.3.self_attn.time_mixer.x_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.4.self_attn.time_mixer.k_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.4.self_attn.time_mixer.x_r": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.5.self_attn.time_mixer.ln_x.bias": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.5.self_attn.time_mixer.x_v": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.6.self_attn.time_mixer.ln_x.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.6.self_attn.time_mixer.x_w": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.7.self_attn.time_mixer.v0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.8.input_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.8.self_attn.time_mixer.w0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.9.post_attention_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.9.self_attn.time_mixer.x_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.1.self_attn.time_mixer.r_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.17.self_attn.time_mixer.r_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.24.self_attn.time_mixer.r_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.7.self_attn.time_mixer.r_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0012_Of_20.pt", | |
| "model.model.layers.11.mlp.up_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.17.mlp.up_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.22.mlp.up_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.3.mlp.up_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.9.mlp.up_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.19.self_attn.time_mixer.value.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.23.self_attn.time_mixer.output.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.27.self_attn.time_mixer.value.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.7.self_attn.time_mixer.output.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.11.self_attn.time_mixer.g2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.15.self_attn.time_mixer.g2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.19.self_attn.time_mixer.g2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.22.self_attn.time_mixer.g2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.26.self_attn.time_mixer.g2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.5.self_attn.time_mixer.g2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.9.self_attn.time_mixer.g2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.1.self_attn.time_mixer.w2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.11.self_attn.time_mixer.w2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.13.self_attn.time_mixer.w2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.15.self_attn.time_mixer.w2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.17.self_attn.time_mixer.w2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.19.self_attn.time_mixer.w2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.20.self_attn.time_mixer.w2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.22.self_attn.time_mixer.w2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.24.self_attn.time_mixer.w2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.26.self_attn.time_mixer.w2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.3.self_attn.time_mixer.w2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.5.self_attn.time_mixer.w2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.7.self_attn.time_mixer.w2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.9.self_attn.time_mixer.w2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.11.self_attn.time_mixer.v2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.15.self_attn.time_mixer.v2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.19.self_attn.time_mixer.v2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.22.self_attn.time_mixer.v2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.26.self_attn.time_mixer.v2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.5.self_attn.time_mixer.v2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.9.self_attn.time_mixer.v2": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.0.self_attn.time_mixer.v0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.1.input_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.1.self_attn.time_mixer.w0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.10.post_attention_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.10.self_attn.time_mixer.x_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.11.self_attn.time_mixer.a0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.11.self_attn.time_mixer.x_g": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.12.self_attn.time_mixer.k_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.12.self_attn.time_mixer.x_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.13.self_attn.time_mixer.k_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.13.self_attn.time_mixer.x_r": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.14.self_attn.time_mixer.ln_x.bias": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.14.self_attn.time_mixer.x_v": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.15.self_attn.time_mixer.ln_x.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.15.self_attn.time_mixer.x_w": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.16.self_attn.time_mixer.v0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.17.input_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.17.self_attn.time_mixer.w0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.18.post_attention_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.18.self_attn.time_mixer.x_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.19.self_attn.time_mixer.a0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.19.self_attn.time_mixer.x_g": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.2.self_attn.time_mixer.k_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.2.self_attn.time_mixer.x_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.20.self_attn.time_mixer.k_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.20.self_attn.time_mixer.x_r": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.21.self_attn.time_mixer.ln_x.bias": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.21.self_attn.time_mixer.x_v": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.22.self_attn.time_mixer.ln_x.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.22.self_attn.time_mixer.x_w": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.23.self_attn.time_mixer.v0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.24.input_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.24.self_attn.time_mixer.w0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.25.post_attention_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.25.self_attn.time_mixer.x_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.26.self_attn.time_mixer.a0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.26.self_attn.time_mixer.x_g": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.27.self_attn.time_mixer.k_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.27.self_attn.time_mixer.x_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.3.self_attn.time_mixer.k_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.3.self_attn.time_mixer.x_r": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.4.self_attn.time_mixer.ln_x.bias": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.4.self_attn.time_mixer.x_v": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.5.self_attn.time_mixer.ln_x.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.5.self_attn.time_mixer.x_w": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.6.self_attn.time_mixer.v0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.7.input_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.7.self_attn.time_mixer.w0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.8.post_attention_layernorm.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.8.self_attn.time_mixer.x_a": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.9.self_attn.time_mixer.a0": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.9.self_attn.time_mixer.x_g": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.10.self_attn.time_mixer.r_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.18.self_attn.time_mixer.r_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.25.self_attn.time_mixer.r_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.8.self_attn.time_mixer.r_k": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0013_Of_20.pt", | |
| "model.model.layers.12.mlp.down_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0014_Of_20.pt", | |
| "model.model.layers.18.mlp.down_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0014_Of_20.pt", | |
| "model.model.layers.23.mlp.down_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0014_Of_20.pt", | |
| "model.model.layers.4.mlp.down_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0014_Of_20.pt", | |
| "model.model.layers.0.self_attn.time_mixer.key.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0014_Of_20.pt", | |
| "model.model.layers.1.self_attn.time_mixer.receptance.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0014_Of_20.pt", | |
| "model.model.layers.11.self_attn.time_mixer.key.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0014_Of_20.pt", | |
| "model.model.layers.12.self_attn.time_mixer.receptance.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0014_Of_20.pt", | |
| "model.model.layers.14.self_attn.time_mixer.key.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0014_Of_20.pt", | |
| "model.model.layers.15.self_attn.time_mixer.receptance.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0014_Of_20.pt", | |
| "model.model.layers.2.self_attn.time_mixer.key.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0014_Of_20.pt", | |
| "model.model.layers.23.self_attn.time_mixer.receptance.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0014_Of_20.pt", | |
| "model.model.layers.3.self_attn.time_mixer.key.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0014_Of_20.pt", | |
| "model.model.layers.7.self_attn.time_mixer.receptance.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0014_Of_20.pt", | |
| "model.model.layers.12.mlp.gate_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0015_Of_20.pt", | |
| "model.model.layers.18.mlp.gate_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0015_Of_20.pt", | |
| "model.model.layers.23.mlp.gate_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0015_Of_20.pt", | |
| "model.model.layers.4.mlp.gate_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0015_Of_20.pt", | |
| "model.model.layers.0.self_attn.time_mixer.output.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0015_Of_20.pt", | |
| "model.model.layers.1.self_attn.time_mixer.value.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0015_Of_20.pt", | |
| "model.model.layers.11.self_attn.time_mixer.output.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0015_Of_20.pt", | |
| "model.model.layers.12.self_attn.time_mixer.value.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0015_Of_20.pt", | |
| "model.model.layers.14.self_attn.time_mixer.output.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0015_Of_20.pt", | |
| "model.model.layers.15.self_attn.time_mixer.value.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0015_Of_20.pt", | |
| "model.model.layers.2.self_attn.time_mixer.output.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0015_Of_20.pt", | |
| "model.model.layers.23.self_attn.time_mixer.value.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0015_Of_20.pt", | |
| "model.model.layers.3.self_attn.time_mixer.output.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0015_Of_20.pt", | |
| "model.model.layers.7.self_attn.time_mixer.value.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0015_Of_20.pt", | |
| "model.model.layers.12.mlp.up_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0016_Of_20.pt", | |
| "model.model.layers.18.mlp.up_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0016_Of_20.pt", | |
| "model.model.layers.23.mlp.up_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0016_Of_20.pt", | |
| "model.model.layers.4.mlp.up_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0016_Of_20.pt", | |
| "model.model.layers.0.self_attn.time_mixer.receptance.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0016_Of_20.pt", | |
| "model.model.layers.10.self_attn.time_mixer.key.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0016_Of_20.pt", | |
| "model.model.layers.11.self_attn.time_mixer.receptance.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0016_Of_20.pt", | |
| "model.model.layers.13.self_attn.time_mixer.key.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0016_Of_20.pt", | |
| "model.model.layers.14.self_attn.time_mixer.receptance.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0016_Of_20.pt", | |
| "model.model.layers.16.self_attn.time_mixer.key.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0016_Of_20.pt", | |
| "model.model.layers.2.self_attn.time_mixer.receptance.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0016_Of_20.pt", | |
| "model.model.layers.24.self_attn.time_mixer.key.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0016_Of_20.pt", | |
| "model.model.layers.3.self_attn.time_mixer.receptance.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0016_Of_20.pt", | |
| "model.model.layers.8.self_attn.time_mixer.key.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0016_Of_20.pt", | |
| "model.model.layers.13.mlp.down_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0017_Of_20.pt", | |
| "model.model.layers.19.mlp.down_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0017_Of_20.pt", | |
| "model.model.layers.24.mlp.down_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0017_Of_20.pt", | |
| "model.model.layers.5.mlp.down_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0017_Of_20.pt", | |
| "model.model.layers.0.self_attn.time_mixer.value.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0017_Of_20.pt", | |
| "model.model.layers.10.self_attn.time_mixer.output.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0017_Of_20.pt", | |
| "model.model.layers.11.self_attn.time_mixer.value.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0017_Of_20.pt", | |
| "model.model.layers.13.self_attn.time_mixer.output.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0017_Of_20.pt", | |
| "model.model.layers.14.self_attn.time_mixer.value.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0017_Of_20.pt", | |
| "model.model.layers.16.self_attn.time_mixer.output.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0017_Of_20.pt", | |
| "model.model.layers.2.self_attn.time_mixer.value.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0017_Of_20.pt", | |
| "model.model.layers.24.self_attn.time_mixer.output.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0017_Of_20.pt", | |
| "model.model.layers.3.self_attn.time_mixer.value.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0017_Of_20.pt", | |
| "model.model.layers.8.self_attn.time_mixer.output.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0017_Of_20.pt", | |
| "model.model.layers.13.mlp.gate_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0018_Of_20.pt", | |
| "model.model.layers.19.mlp.gate_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0018_Of_20.pt", | |
| "model.model.layers.24.mlp.gate_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0018_Of_20.pt", | |
| "model.model.layers.5.mlp.gate_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0018_Of_20.pt", | |
| "model.model.layers.1.self_attn.time_mixer.key.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0018_Of_20.pt", | |
| "model.model.layers.10.self_attn.time_mixer.receptance.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0018_Of_20.pt", | |
| "model.model.layers.12.self_attn.time_mixer.key.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0018_Of_20.pt", | |
| "model.model.layers.13.self_attn.time_mixer.receptance.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0018_Of_20.pt", | |
| "model.model.layers.15.self_attn.time_mixer.key.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0018_Of_20.pt", | |
| "model.model.layers.16.self_attn.time_mixer.receptance.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0018_Of_20.pt", | |
| "model.model.layers.20.self_attn.time_mixer.key.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0018_Of_20.pt", | |
| "model.model.layers.24.self_attn.time_mixer.receptance.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0018_Of_20.pt", | |
| "model.model.layers.4.self_attn.time_mixer.key.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0018_Of_20.pt", | |
| "model.model.layers.8.self_attn.time_mixer.receptance.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0018_Of_20.pt", | |
| "model.model.layers.13.mlp.up_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0019_Of_20.pt", | |
| "model.model.layers.19.mlp.up_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0019_Of_20.pt", | |
| "model.model.layers.24.mlp.up_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0019_Of_20.pt", | |
| "model.model.layers.5.mlp.up_proj.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0019_Of_20.pt", | |
| "model.model.layers.1.self_attn.time_mixer.output.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0019_Of_20.pt", | |
| "model.model.layers.10.self_attn.time_mixer.value.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0019_Of_20.pt", | |
| "model.model.layers.12.self_attn.time_mixer.output.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0019_Of_20.pt", | |
| "model.model.layers.13.self_attn.time_mixer.value.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0019_Of_20.pt", | |
| "model.model.layers.15.self_attn.time_mixer.output.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0019_Of_20.pt", | |
| "model.model.layers.16.self_attn.time_mixer.value.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0019_Of_20.pt", | |
| "model.model.layers.20.self_attn.time_mixer.output.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0019_Of_20.pt", | |
| "model.model.layers.24.self_attn.time_mixer.value.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0019_Of_20.pt", | |
| "model.model.layers.4.self_attn.time_mixer.output.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0019_Of_20.pt", | |
| "model.model.layers.8.self_attn.time_mixer.value.weight": "/home/yueyulin/model/qwen_7b_distill_stage2_splits/7b_stage2_model_converted_0019_Of_20.pt" | |
| } |