Upload Qwen2ForCausalLM

config.json

@@ -17,9 +17,9 @@
   "num_hidden_layers": 24,
   "num_key_value_heads": 2,
   "quantization_config": {
-    "_load_in_4bit": false,
+    "_load_in_4bit": true,
     "_load_in_8bit": false,
-    "bnb_4bit_compute_dtype": "float32",
+    "bnb_4bit_compute_dtype": "float16",
     "bnb_4bit_quant_storage": "uint8",
     "bnb_4bit_quant_type": "fp4",
     "bnb_4bit_use_double_quant": false,
@@ -27,7 +27,7 @@
     "llm_int8_has_fp16_weight": false,
     "llm_int8_skip_modules": null,
     "llm_int8_threshold": 6.0,
-    "load_in_4bit": false,
+    "load_in_4bit": true,
     "load_in_8bit": false,
     "quant_method": "bitsandbytes"
   },