llama_model_loader: loaded meta data with 33 key-value pairs and 724 tensors from Hermes-3-Llama-3.1-70B-IMat-GGUF/Hermes-3-Llama-3.1-70B.Q8_0.gguf.hardlink.gguf (version GGUF V3 (latest)) llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output. llama_model_loader: - kv 0: general.architecture str = llama llama_model_loader: - kv 1: general.type str = model llama_model_loader: - kv 2: general.name str = Hermes 3 Llama 3.1 70B llama_model_loader: - kv 3: general.basename str = Hermes-3-Llama-3.1 llama_model_loader: - kv 4: general.size_label str = 70B llama_model_loader: - kv 5: general.license str = llama3 llama_model_loader: - kv 6: general.base_model.count u32 = 1 llama_model_loader: - kv 7: general.base_model.0.name str = Meta Llama 3.1 70B llama_model_loader: - kv 8: general.base_model.0.organization str = Meta Llama llama_model_loader: - kv 9: general.base_model.0.repo_url str = https://huggingface.co/meta-llama/Met... llama_model_loader: - kv 10: general.tags arr[str,12] = ["Llama-3", "instruct", "finetune", "... llama_model_loader: - kv 11: general.languages arr[str,1] = ["en"] llama_model_loader: - kv 12: llama.block_count u32 = 80 llama_model_loader: - kv 13: llama.context_length u32 = 131072 llama_model_loader: - kv 14: llama.embedding_length u32 = 8192 llama_model_loader: - kv 15: llama.feed_forward_length u32 = 28672 llama_model_loader: - kv 16: llama.attention.head_count u32 = 64 llama_model_loader: - kv 17: llama.attention.head_count_kv u32 = 8 llama_model_loader: - kv 18: llama.rope.freq_base f32 = 500000.000000 llama_model_loader: - kv 19: llama.attention.layer_norm_rms_epsilon f32 = 0.000010 llama_model_loader: - kv 20: general.file_type u32 = 7 llama_model_loader: - kv 21: llama.vocab_size u32 = 128256 llama_model_loader: - kv 22: llama.rope.dimension_count u32 = 128 llama_model_loader: - kv 23: tokenizer.ggml.model str = gpt2 llama_model_loader: - kv 24: tokenizer.ggml.pre str = llama-bpe llama_model_loader: - kv 25: tokenizer.ggml.tokens arr[str,128256] = ["!", "\"", "#", "$", "%", "&", "'", ... llama_model_loader: - kv 26: tokenizer.ggml.token_type arr[i32,128256] = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ... llama_model_loader: - kv 27: tokenizer.ggml.merges arr[str,280147] = ["Ġ Ġ", "Ġ ĠĠĠ", "ĠĠ ĠĠ", "... llama_model_loader: - kv 28: tokenizer.ggml.bos_token_id u32 = 128000 llama_model_loader: - kv 29: tokenizer.ggml.eos_token_id u32 = 128039 llama_model_loader: - kv 30: tokenizer.ggml.padding_token_id u32 = 128001 llama_model_loader: - kv 31: tokenizer.chat_template str = {{bos_token}}{% for message in messag... llama_model_loader: - kv 32: general.quantization_version u32 = 2 llama_model_loader: - type f32: 162 tensors llama_model_loader: - type q8_0: 562 tensors llm_load_vocab: special tokens cache size = 256 llm_load_vocab: token to piece cache size = 0.7994 MB llm_load_print_meta: format = GGUF V3 (latest) llm_load_print_meta: arch = llama llm_load_print_meta: vocab type = BPE llm_load_print_meta: n_vocab = 128256 llm_load_print_meta: n_merges = 280147 llm_load_print_meta: vocab_only = 0 llm_load_print_meta: n_ctx_train = 131072 llm_load_print_meta: n_embd = 8192 llm_load_print_meta: n_layer = 80 llm_load_print_meta: n_head = 64 llm_load_print_meta: n_head_kv = 8 llm_load_print_meta: n_rot = 128 llm_load_print_meta: n_swa = 0 llm_load_print_meta: n_embd_head_k = 128 llm_load_print_meta: n_embd_head_v = 128 llm_load_print_meta: n_gqa = 8 llm_load_print_meta: n_embd_k_gqa = 1024 llm_load_print_meta: n_embd_v_gqa = 1024 llm_load_print_meta: f_norm_eps = 0.0e+00 llm_load_print_meta: f_norm_rms_eps = 1.0e-05 llm_load_print_meta: f_clamp_kqv = 0.0e+00 llm_load_print_meta: f_max_alibi_bias = 0.0e+00 llm_load_print_meta: f_logit_scale = 0.0e+00 llm_load_print_meta: n_ff = 28672 llm_load_print_meta: n_expert = 0 llm_load_print_meta: n_expert_used = 0 llm_load_print_meta: causal attn = 1 llm_load_print_meta: pooling type = 0 llm_load_print_meta: rope type = 0 llm_load_print_meta: rope scaling = linear llm_load_print_meta: freq_base_train = 500000.0 llm_load_print_meta: freq_scale_train = 1 llm_load_print_meta: n_ctx_orig_yarn = 131072 llm_load_print_meta: rope_finetuned = unknown llm_load_print_meta: ssm_d_conv = 0 llm_load_print_meta: ssm_d_inner = 0 llm_load_print_meta: ssm_d_state = 0 llm_load_print_meta: ssm_dt_rank = 0 llm_load_print_meta: model type = 70B llm_load_print_meta: model ftype = Q8_0 llm_load_print_meta: model params = 70.55 B llm_load_print_meta: model size = 69.82 GiB (8.50 BPW) llm_load_print_meta: general.name = Hermes 3 Llama 3.1 70B llm_load_print_meta: BOS token = 128000 '<|begin_of_text|>' llm_load_print_meta: EOS token = 128039 '<|im_end|>' llm_load_print_meta: PAD token = 128001 '<|end_of_text|>' llm_load_print_meta: LF token = 128 'Ä' llm_load_print_meta: EOT token = 128039 '<|im_end|>' llm_load_print_meta: max token length = 256 ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no ggml_cuda_init: found 1 CUDA devices: Device 0: NVIDIA GeForce RTX 4090, compute capability 8.9, VMM: yes llm_load_tensors: ggml ctx size = 0.68 MiB llm_load_tensors: offloading 25 repeating layers to GPU llm_load_tensors: offloaded 25/81 layers to GPU llm_load_tensors: CPU buffer size = 71494.28 MiB llm_load_tensors: CUDA0 buffer size = 21676.57 MiB .................................................................................................... llama_new_context_with_model: n_ctx = 512 llama_new_context_with_model: n_batch = 512 llama_new_context_with_model: n_ubatch = 512 llama_new_context_with_model: flash_attn = 0 llama_new_context_with_model: freq_base = 500000.0 llama_new_context_with_model: freq_scale = 1 llama_kv_cache_init: CUDA_Host KV buffer size = 110.00 MiB llama_kv_cache_init: CUDA0 KV buffer size = 50.00 MiB llama_new_context_with_model: KV self size = 160.00 MiB, K (f16): 80.00 MiB, V (f16): 80.00 MiB llama_new_context_with_model: CUDA_Host output buffer size = 0.49 MiB llama_new_context_with_model: CUDA0 compute buffer size = 1331.12 MiB llama_new_context_with_model: CUDA_Host compute buffer size = 17.01 MiB llama_new_context_with_model: graph nodes = 2566 llama_new_context_with_model: graph splits = 719 system_info: n_threads = 25 / 32 | AVX = 1 | AVX_VNNI = 0 | AVX2 = 1 | AVX512 = 1 | AVX512_VBMI = 1 | AVX512_VNNI = 1 | AVX512_BF16 = 1 | FMA = 1 | NEON = 0 | SVE = 0 | ARM_FMA = 0 | F16C = 1 | FP16_VA = 0 | WASM_SIMD = 0 | BLAS = 1 | SSE3 = 1 | SSSE3 = 1 | VSX = 0 | MATMUL_INT8 = 0 | LLAMAFILE = 1 | compute_imatrix: tokenizing the input .. compute_imatrix: tokenization took 42.062 ms compute_imatrix: computing over 125 chunks with batch_size 512 compute_imatrix: 6.17 seconds per pass - ETA 12.85 minutes [1]5.5350,[2]4.3772,[3]3.8197,[4]4.6004,[5]4.6317,[6]3.9145,[7]3.8311,[8]4.2072,[9]4.4117, save_imatrix: stored collected data after 10 chunks in Hermes-3-Llama-3.1-70B-IMat-GGUF/imatrix.dat [10]4.0759,[11]4.4896,[12]4.9017,[13]5.2563,[14]5.6161,[15]5.8285,[16]6.1324,[17]6.2960,[18]6.0750,[19]5.8143, save_imatrix: stored collected data after 20 chunks in Hermes-3-Llama-3.1-70B-IMat-GGUF/imatrix.dat [20]5.8144,[21]5.8849,[22]5.8756,[23]6.0320,[24]6.0210,[25]6.3261,[26]6.3191,[27]5.9113,[28]5.5892,[29]5.5933, save_imatrix: stored collected data after 30 chunks in Hermes-3-Llama-3.1-70B-IMat-GGUF/imatrix.dat [30]5.5664,[31]5.3252,[32]5.1001,[33]4.9893,[34]4.9118,[35]4.9732,[36]5.0064,[37]4.9847,[38]5.0464,[39]5.1761, save_imatrix: stored collected data after 40 chunks in Hermes-3-Llama-3.1-70B-IMat-GGUF/imatrix.dat [40]5.2587,[41]5.0658,[42]4.8822,[43]4.7117,[44]4.5586,[45]4.4981,[46]4.4530,[47]4.5595,[48]4.6451,[49]4.7463, save_imatrix: stored collected data after 50 chunks in Hermes-3-Llama-3.1-70B-IMat-GGUF/imatrix.dat [50]4.6968,[51]4.7996,[52]4.9034,[53]4.9806,[54]5.0345,[55]5.0484,[56]5.0379,[57]5.0703,[58]5.0689,[59]5.0908, save_imatrix: stored collected data after 60 chunks in Hermes-3-Llama-3.1-70B-IMat-GGUF/imatrix.dat [60]5.0870,[61]5.0991,[62]5.1406,[63]5.1984,[64]5.1646,[65]5.1586,[66]5.1768,[67]5.1865,[68]5.1912,[69]5.1962, save_imatrix: stored collected data after 70 chunks in Hermes-3-Llama-3.1-70B-IMat-GGUF/imatrix.dat [70]5.2210,[71]5.2371,[72]5.2564,[73]5.2562,[74]5.2432,[75]5.2579,[76]5.2687,[77]5.2661,[78]5.2628,[79]5.3063, save_imatrix: stored collected data after 80 chunks in Hermes-3-Llama-3.1-70B-IMat-GGUF/imatrix.dat [80]5.3389,[81]5.3465,[82]5.3624,[83]5.3993,[84]5.3458,[85]5.3456,[86]5.3547,[87]5.3843,[88]5.4268,[89]5.4569, save_imatrix: stored collected data after 90 chunks in Hermes-3-Llama-3.1-70B-IMat-GGUF/imatrix.dat [90]5.4428,[91]5.4191,[92]5.3993,[93]5.3862,[94]5.3646,[95]5.3502,[96]5.3324,[97]5.3531,[98]5.3941,[99]5.4577, save_imatrix: stored collected data after 100 chunks in Hermes-3-Llama-3.1-70B-IMat-GGUF/imatrix.dat [100]5.5153,[101]5.5593,[102]5.6484,[103]5.6790,[104]5.7133,[105]5.6708,[106]5.6876,[107]5.6561,[108]5.5959,[109]5.5353, save_imatrix: stored collected data after 110 chunks in Hermes-3-Llama-3.1-70B-IMat-GGUF/imatrix.dat [110]5.5757,[111]5.6069,[112]5.6253,[113]5.6315,[114]5.6624,[115]5.7002,[116]5.7168,[117]5.7379,[118]5.7737,[119]5.7478, save_imatrix: stored collected data after 120 chunks in Hermes-3-Llama-3.1-70B-IMat-GGUF/imatrix.dat [120]5.6694,[121]5.5910,[122]5.5181,[123]5.4504,[124]5.3941,[125]5.3246, save_imatrix: stored collected data after 125 chunks in Hermes-3-Llama-3.1-70B-IMat-GGUF/imatrix.dat llama_print_timings: load time = 32580.75 ms llama_print_timings: sample time = 0.00 ms / 1 runs ( 0.00 ms per token, inf tokens per second) llama_print_timings: prompt eval time = 723343.07 ms / 64000 tokens ( 11.30 ms per token, 88.48 tokens per second) llama_print_timings: eval time = 0.00 ms / 1 runs ( 0.00 ms per token, inf tokens per second) llama_print_timings: total time = 751504.95 ms / 64001 tokens Final estimate: PPL = 5.3246 +/- 0.07650