llama_model_loader: loaded meta data with 51 key-value pairs and 959 tensors from DeepSeek-Coder-V2-Instruct-0724-IMat-GGUF/DeepSeek-Coder-V2-Instruct-0724.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 = deepseek2 llama_model_loader: - kv 1: general.type str = model llama_model_loader: - kv 2: general.name str = DeepSeek Coder V2 Instruct 0724 llama_model_loader: - kv 3: general.version str = 0724 llama_model_loader: - kv 4: general.finetune str = Instruct llama_model_loader: - kv 5: general.basename str = DeepSeek-Coder-V2 llama_model_loader: - kv 6: general.size_label str = 160x14B llama_model_loader: - kv 7: general.license str = other llama_model_loader: - kv 8: general.license.name str = deepseek-license llama_model_loader: - kv 9: general.license.link str = LICENSE llama_model_loader: - kv 10: general.base_model.count u32 = 1 llama_model_loader: - kv 11: general.base_model.0.name str = DeepSeek Coder V2 Base llama_model_loader: - kv 12: general.base_model.0.organization str = Deepseek Ai llama_model_loader: - kv 13: general.base_model.0.repo_url str = https://huggingface.co./deepseek-ai/De... llama_model_loader: - kv 14: deepseek2.block_count u32 = 60 llama_model_loader: - kv 15: deepseek2.context_length u32 = 163840 llama_model_loader: - kv 16: deepseek2.embedding_length u32 = 5120 llama_model_loader: - kv 17: deepseek2.feed_forward_length u32 = 12288 llama_model_loader: - kv 18: deepseek2.attention.head_count u32 = 128 llama_model_loader: - kv 19: deepseek2.attention.head_count_kv u32 = 128 llama_model_loader: - kv 20: deepseek2.rope.freq_base f32 = 10000.000000 llama_model_loader: - kv 21: deepseek2.attention.layer_norm_rms_epsilon f32 = 0.000001 llama_model_loader: - kv 22: deepseek2.expert_used_count u32 = 6 llama_model_loader: - kv 23: general.file_type u32 = 7 llama_model_loader: - kv 24: deepseek2.leading_dense_block_count u32 = 1 llama_model_loader: - kv 25: deepseek2.vocab_size u32 = 102400 llama_model_loader: - kv 26: deepseek2.attention.q_lora_rank u32 = 1536 llama_model_loader: - kv 27: deepseek2.attention.kv_lora_rank u32 = 512 llama_model_loader: - kv 28: deepseek2.attention.key_length u32 = 192 llama_model_loader: - kv 29: deepseek2.attention.value_length u32 = 128 llama_model_loader: - kv 30: deepseek2.expert_feed_forward_length u32 = 1536 llama_model_loader: - kv 31: deepseek2.expert_count u32 = 160 llama_model_loader: - kv 32: deepseek2.expert_shared_count u32 = 2 llama_model_loader: - kv 33: deepseek2.expert_weights_scale f32 = 16.000000 llama_model_loader: - kv 34: deepseek2.rope.dimension_count u32 = 64 llama_model_loader: - kv 35: deepseek2.rope.scaling.type str = yarn llama_model_loader: - kv 36: deepseek2.rope.scaling.factor f32 = 40.000000 llama_model_loader: - kv 37: deepseek2.rope.scaling.original_context_length u32 = 4096 llama_model_loader: - kv 38: deepseek2.rope.scaling.yarn_log_multiplier f32 = 0.100000 llama_model_loader: - kv 39: tokenizer.ggml.model str = gpt2 llama_model_loader: - kv 40: tokenizer.ggml.pre str = deepseek-llm llama_model_loader: - kv 41: tokenizer.ggml.tokens arr[str,102400] = ["!", "\"", "#", "$", "%", "&", "'", ... llama_model_loader: - kv 42: tokenizer.ggml.token_type arr[i32,102400] = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ... llama_model_loader: - kv 43: tokenizer.ggml.merges arr[str,99757] = ["Ġ Ġ", "Ġ t", "Ġ a", "i n", "h e... llama_model_loader: - kv 44: tokenizer.ggml.bos_token_id u32 = 100000 llama_model_loader: - kv 45: tokenizer.ggml.eos_token_id u32 = 100001 llama_model_loader: - kv 46: tokenizer.ggml.padding_token_id u32 = 100001 llama_model_loader: - kv 47: tokenizer.ggml.add_bos_token bool = true llama_model_loader: - kv 48: tokenizer.ggml.add_eos_token bool = false llama_model_loader: - kv 49: tokenizer.chat_template str = {% if not add_generation_prompt is de... llama_model_loader: - kv 50: general.quantization_version u32 = 2 llama_model_loader: - type f32: 300 tensors llama_model_loader: - type q8_0: 659 tensors llm_load_vocab: special tokens cache size = 18 llm_load_vocab: token to piece cache size = 0.6411 MB llm_load_print_meta: format = GGUF V3 (latest) llm_load_print_meta: arch = deepseek2 llm_load_print_meta: vocab type = BPE llm_load_print_meta: n_vocab = 102400 llm_load_print_meta: n_merges = 99757 llm_load_print_meta: vocab_only = 0 llm_load_print_meta: n_ctx_train = 163840 llm_load_print_meta: n_embd = 5120 llm_load_print_meta: n_layer = 60 llm_load_print_meta: n_head = 128 llm_load_print_meta: n_head_kv = 128 llm_load_print_meta: n_rot = 64 llm_load_print_meta: n_swa = 0 llm_load_print_meta: n_embd_head_k = 192 llm_load_print_meta: n_embd_head_v = 128 llm_load_print_meta: n_gqa = 1 llm_load_print_meta: n_embd_k_gqa = 24576 llm_load_print_meta: n_embd_v_gqa = 16384 llm_load_print_meta: f_norm_eps = 0.0e+00 llm_load_print_meta: f_norm_rms_eps = 1.0e-06 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 = 12288 llm_load_print_meta: n_expert = 160 llm_load_print_meta: n_expert_used = 6 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 = yarn llm_load_print_meta: freq_base_train = 10000.0 llm_load_print_meta: freq_scale_train = 0.025 llm_load_print_meta: n_ctx_orig_yarn = 4096 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: ssm_dt_b_c_rms = 0 llm_load_print_meta: model type = 236B llm_load_print_meta: model ftype = Q8_0 llm_load_print_meta: model params = 235.74 B llm_load_print_meta: model size = 233.41 GiB (8.50 BPW) llm_load_print_meta: general.name = DeepSeek Coder V2 Instruct 0724 llm_load_print_meta: BOS token = 100000 '<｜begin▁of▁sentence｜>' llm_load_print_meta: EOS token = 100001 '<｜end▁of▁sentence｜>' llm_load_print_meta: PAD token = 100001 '<｜end▁of▁sentence｜>' llm_load_print_meta: LF token = 126 'Ä' llm_load_print_meta: max token length = 256 llm_load_print_meta: n_layer_dense_lead = 1 llm_load_print_meta: n_lora_q = 1536 llm_load_print_meta: n_lora_kv = 512 llm_load_print_meta: n_ff_exp = 1536 llm_load_print_meta: n_expert_shared = 2 llm_load_print_meta: expert_weights_scale = 16.0 llm_load_print_meta: rope_yarn_log_mul = 0.1000 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 = 0.80 MiB llm_load_tensors: offloading 5 repeating layers to GPU llm_load_tensors: offloaded 5/61 layers to GPU llm_load_tensors: CPU buffer size = 218873.36 MiB llm_load_tensors: CUDA0 buffer size = 20135.96 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 = 10000.0 llama_new_context_with_model: freq_scale = 0.025 llama_kv_cache_init: CUDA_Host KV buffer size = 2200.00 MiB llama_kv_cache_init: CUDA0 KV buffer size = 200.00 MiB llama_new_context_with_model: KV self size = 2400.00 MiB, K (f16): 1440.00 MiB, V (f16): 960.00 MiB llama_new_context_with_model: CUDA_Host output buffer size = 0.39 MiB llama_new_context_with_model: CUDA0 compute buffer size = 1422.00 MiB llama_new_context_with_model: CUDA_Host compute buffer size = 81.01 MiB llama_new_context_with_model: graph nodes = 4480 llama_new_context_with_model: graph splits = 990 system_info: n_threads = 25 (n_threads_batch = 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 526.799 ms compute_imatrix: computing over 139 chunks with batch_size 512 compute_imatrix: 90.65 seconds per pass - ETA 3 hours 30.00 minutes [1]5.7772,[2]3.8835,[3]3.8059,[4]4.1672,[5]4.1086,[6]3.9026,[7]4.1099,[8]3.9733,[9]4.4323, save_imatrix: stored collected data after 10 chunks in DeepSeek-Coder-V2-Instruct-0724-IMat-GGUF/imatrix.dat [10]4.6201,[11]4.3030,[12]4.5590,[13]4.8871,[14]5.1335,[15]5.2601,[16]5.4875,[17]5.6677,[18]5.7927,[19]5.8962, save_imatrix: stored collected data after 20 chunks in DeepSeek-Coder-V2-Instruct-0724-IMat-GGUF/imatrix.dat [20]5.6232,[21]5.6381,[22]5.5802,[23]5.6274,[24]5.5363,[25]5.6976,[26]5.6137,[27]5.7519,[28]5.6031,[29]5.3284, save_imatrix: stored collected data after 30 chunks in DeepSeek-Coder-V2-Instruct-0724-IMat-GGUF/imatrix.dat [30]5.1049,[31]5.0415,[32]4.9882,[33]4.8558,[34]4.6419,[35]4.4657,[36]4.4172,[37]4.3647,[38]4.3780,[39]4.3155, save_imatrix: stored collected data after 40 chunks in DeepSeek-Coder-V2-Instruct-0724-IMat-GGUF/imatrix.dat [40]4.2715,[41]4.2049,[42]4.1509,[43]4.1936,[44]4.2619,[45]4.3487,[46]4.3656,[47]4.2404,[48]4.1211,[49]4.0191, save_imatrix: stored collected data after 50 chunks in DeepSeek-Coder-V2-Instruct-0724-IMat-GGUF/imatrix.dat [50]3.9192,[51]3.9238,[52]3.8966,[53]3.9838,[54]4.0627,[55]4.1495,[56]4.1101,[57]4.1283,[58]4.1489,[59]4.2108, save_imatrix: stored collected data after 60 chunks in DeepSeek-Coder-V2-Instruct-0724-IMat-GGUF/imatrix.dat [60]4.2794,[61]4.3451,[62]4.3897,[63]4.4198,[64]4.4604,[65]4.4913,[66]4.4892,[67]4.4981,[68]4.5029,[69]4.5437, save_imatrix: stored collected data after 70 chunks in DeepSeek-Coder-V2-Instruct-0724-IMat-GGUF/imatrix.dat [70]4.5644,[71]4.5709,[72]4.6022,[73]4.6110,[74]4.6151,[75]4.6210,[76]4.6322,[77]4.6445,[78]4.6694,[79]4.6579, save_imatrix: stored collected data after 80 chunks in DeepSeek-Coder-V2-Instruct-0724-IMat-GGUF/imatrix.dat [80]4.6792,[81]4.6795,[82]4.6854,[83]4.6823,[84]4.6838,[85]4.6807,[86]4.6792,[87]4.6723,[88]4.6972,[89]4.7218, save_imatrix: stored collected data after 90 chunks in DeepSeek-Coder-V2-Instruct-0724-IMat-GGUF/imatrix.dat [90]4.7272,[91]4.7600,[92]4.7942,[93]4.7672,[94]4.7706,[95]4.7531,[96]4.7771,[97]4.7976,[98]4.7951,[99]4.7560, save_imatrix: stored collected data after 100 chunks in DeepSeek-Coder-V2-Instruct-0724-IMat-GGUF/imatrix.dat [100]4.7169,[101]4.6779,[102]4.6376,[103]4.5994,[104]4.5669,[105]4.5309,[106]4.4965,[107]4.4640,[108]4.4398,[109]4.4545, save_imatrix: stored collected data after 110 chunks in DeepSeek-Coder-V2-Instruct-0724-IMat-GGUF/imatrix.dat [110]4.4866,[111]4.5306,[112]4.5749,[113]4.6178,[114]4.6884,[115]4.7267,[116]4.7498,[117]4.7678,[118]4.7937,[119]4.7927, save_imatrix: stored collected data after 120 chunks in DeepSeek-Coder-V2-Instruct-0724-IMat-GGUF/imatrix.dat [120]4.7703,[121]4.7169,[122]4.6675,[123]4.6978,[124]4.7296,[125]4.7339,[126]4.7399,[127]4.7519,[128]4.7785,[129]4.7832, save_imatrix: stored collected data after 130 chunks in DeepSeek-Coder-V2-Instruct-0724-IMat-GGUF/imatrix.dat [130]4.7964,[131]4.8182,[132]4.8148,[133]4.8068,[134]4.7571,[135]4.7070,[136]4.7102,[137]4.6635,[138]4.6222,[139]4.5770, save_imatrix: stored collected data after 139 chunks in DeepSeek-Coder-V2-Instruct-0724-IMat-GGUF/imatrix.dat llama_perf_print: load time = 194329.99 ms llama_perf_print: prompt eval time = 15123291.91 ms / 71168 tokens ( 212.50 ms per token, 4.71 tokens per second) llama_perf_print: eval time = 0.00 ms / 1 runs ( 0.00 ms per token, inf tokens per second) llama_perf_print: total time = 15252664.04 ms / 71169 tokens Final estimate: PPL = 4.5770 +/- 0.05779