llama_model_loader: loaded meta data with 23 key-value pairs and 479 tensors from Qwen2-57B-A14B-Instruct-IMat-GGUF/Qwen2-57B-A14B-Instruct.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 = qwen2moe llama_model_loader: - kv 1: general.name str = Qwen2-57B-A14B-Instruct llama_model_loader: - kv 2: qwen2moe.block_count u32 = 28 llama_model_loader: - kv 3: qwen2moe.context_length u32 = 32768 llama_model_loader: - kv 4: qwen2moe.embedding_length u32 = 3584 llama_model_loader: - kv 5: qwen2moe.feed_forward_length u32 = 20480 llama_model_loader: - kv 6: qwen2moe.attention.head_count u32 = 28 llama_model_loader: - kv 7: qwen2moe.attention.head_count_kv u32 = 4 llama_model_loader: - kv 8: qwen2moe.rope.freq_base f32 = 1000000.000000 llama_model_loader: - kv 9: qwen2moe.attention.layer_norm_rms_epsilon f32 = 0.000001 llama_model_loader: - kv 10: qwen2moe.expert_used_count u32 = 8 llama_model_loader: - kv 11: general.file_type u32 = 7 llama_model_loader: - kv 12: qwen2moe.expert_count u32 = 64 llama_model_loader: - kv 13: tokenizer.ggml.model str = gpt2 llama_model_loader: - kv 14: tokenizer.ggml.pre str = qwen2 llama_model_loader: - kv 15: tokenizer.ggml.tokens arr[str,151936] = ["!", "\"", "#", "$", "%", "&", "'", ... llama_model_loader: - kv 16: tokenizer.ggml.token_type arr[i32,151936] = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ... llama_model_loader: - kv 17: tokenizer.ggml.merges arr[str,151387] = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "Ġ t",... llama_model_loader: - kv 18: tokenizer.ggml.eos_token_id u32 = 151645 llama_model_loader: - kv 19: tokenizer.ggml.padding_token_id u32 = 151643 llama_model_loader: - kv 20: tokenizer.ggml.bos_token_id u32 = 151643 llama_model_loader: - kv 21: tokenizer.chat_template str = {% for message in messages %}{% if lo... llama_model_loader: - kv 22: general.quantization_version u32 = 2 llama_model_loader: - type f32: 197 tensors llama_model_loader: - type q8_0: 282 tensors llm_load_vocab: special tokens cache size = 293 llm_load_vocab: token to piece cache size = 0.9338 MB llm_load_print_meta: format = GGUF V3 (latest) llm_load_print_meta: arch = qwen2moe llm_load_print_meta: vocab type = BPE llm_load_print_meta: n_vocab = 151936 llm_load_print_meta: n_merges = 151387 llm_load_print_meta: n_ctx_train = 32768 llm_load_print_meta: n_embd = 3584 llm_load_print_meta: n_head = 28 llm_load_print_meta: n_head_kv = 4 llm_load_print_meta: n_layer = 28 llm_load_print_meta: n_rot = 128 llm_load_print_meta: n_embd_head_k = 128 llm_load_print_meta: n_embd_head_v = 128 llm_load_print_meta: n_gqa = 7 llm_load_print_meta: n_embd_k_gqa = 512 llm_load_print_meta: n_embd_v_gqa = 512 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 = 20480 llm_load_print_meta: n_expert = 64 llm_load_print_meta: n_expert_used = 8 llm_load_print_meta: causal attn = 1 llm_load_print_meta: pooling type = 0 llm_load_print_meta: rope type = 2 llm_load_print_meta: rope scaling = linear llm_load_print_meta: freq_base_train = 1000000.0 llm_load_print_meta: freq_scale_train = 1 llm_load_print_meta: n_ctx_orig_yarn = 32768 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 = ?B llm_load_print_meta: model ftype = Q8_0 llm_load_print_meta: model params = 57.41 B llm_load_print_meta: model size = 56.83 GiB (8.50 BPW) llm_load_print_meta: general.name = Qwen2-57B-A14B-Instruct llm_load_print_meta: BOS token = 151643 '<|endoftext|>' llm_load_print_meta: EOS token = 151645 '<|im_end|>' llm_load_print_meta: PAD token = 151643 '<|endoftext|>' llm_load_print_meta: LF token = 148848 'ÄĬ' llm_load_print_meta: EOT token = 151645 '<|im_end|>' ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no ggml_cuda_init: CUDA_USE_TENSOR_CORES: yes 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 = 0.43 MiB llm_load_tensors: offloading 10 repeating layers to GPU llm_load_tensors: offloaded 10/29 layers to GPU llm_load_tensors: CPU buffer size = 58190.19 MiB llm_load_tensors: CUDA0 buffer size = 20388.09 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 = 1000000.0 llama_new_context_with_model: freq_scale = 1 llama_kv_cache_init: CUDA_Host KV buffer size = 18.00 MiB llama_kv_cache_init: CUDA0 KV buffer size = 10.00 MiB llama_new_context_with_model: KV self size = 28.00 MiB, K (f16): 14.00 MiB, V (f16): 14.00 MiB llama_new_context_with_model: CUDA_Host output buffer size = 0.58 MiB llama_new_context_with_model: CUDA0 compute buffer size = 862.52 MiB llama_new_context_with_model: CUDA_Host compute buffer size = 8.01 MiB llama_new_context_with_model: graph nodes = 1910 llama_new_context_with_model: graph splits = 346 system_info: n_threads = 32 / 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 182.937 ms compute_imatrix: computing over 128 chunks with batch_size 512 compute_imatrix: 4.26 seconds per pass - ETA 9.08 minutes [1]4.9575,[2]3.4740,[3]3.1781,[4]3.5055,[5]3.4223,[6]3.1119,[7]3.2700,[8]3.2929,[9]3.7110, save_imatrix: stored collected data after 10 chunks in Qwen2-57B-A14B-Instruct-IMat-GGUF/imatrix.dat [10]3.6668,[11]3.6189,[12]3.9895,[13]4.4287,[14]4.6725,[15]5.0404,[16]5.3116,[17]5.4825,[18]5.6501,[19]5.4929, save_imatrix: stored collected data after 20 chunks in Qwen2-57B-A14B-Instruct-IMat-GGUF/imatrix.dat [20]5.6098,[21]5.6066,[22]5.6538,[23]5.6008,[24]5.8175,[25]5.9804,[26]5.8777,[27]5.9822,[28]6.0379,[29]6.1864, save_imatrix: stored collected data after 30 chunks in Qwen2-57B-A14B-Instruct-IMat-GGUF/imatrix.dat [30]6.1506,[31]5.9725,[32]5.7566,[33]5.6509,[34]5.5608,[35]5.5154,[36]5.5215,[37]5.5629,[38]5.6044,[39]5.5722, save_imatrix: stored collected data after 40 chunks in Qwen2-57B-A14B-Instruct-IMat-GGUF/imatrix.dat [40]5.6990,[41]5.7423,[42]5.9444,[43]6.1088,[44]6.2754,[45]6.4090,[46]6.5149,[47]6.4206,[48]6.4557,[49]6.5417, save_imatrix: stored collected data after 50 chunks in Qwen2-57B-A14B-Instruct-IMat-GGUF/imatrix.dat [50]6.6096,[51]6.5266,[52]6.5961,[53]6.7232,[54]6.8086,[55]6.8662,[56]6.9126,[57]6.9567,[58]6.9946,[59]7.0255, save_imatrix: stored collected data after 60 chunks in Qwen2-57B-A14B-Instruct-IMat-GGUF/imatrix.dat [60]7.0625,[61]7.0266,[62]6.9920,[63]7.0473,[64]7.1158,[65]7.0680,[66]7.0670,[67]7.0863,[68]7.0229,[69]6.9689, save_imatrix: stored collected data after 70 chunks in Qwen2-57B-A14B-Instruct-IMat-GGUF/imatrix.dat [70]6.9623,[71]6.9463,[72]6.9431,[73]6.9588,[74]6.9001,[75]6.8524,[76]6.8061,[77]6.7868,[78]6.7740,[79]6.7511, save_imatrix: stored collected data after 80 chunks in Qwen2-57B-A14B-Instruct-IMat-GGUF/imatrix.dat [80]6.6806,[81]6.7077,[82]6.7016,[83]6.6616,[84]6.6788,[85]6.6970,[86]6.6610,[87]6.6396,[88]6.6213,[89]6.6418, save_imatrix: stored collected data after 90 chunks in Qwen2-57B-A14B-Instruct-IMat-GGUF/imatrix.dat [90]6.6657,[91]6.6567,[92]6.6193,[93]6.5897,[94]6.5406,[95]6.5056,[96]6.4673,[97]6.4356,[98]6.3961,[99]6.3680, save_imatrix: stored collected data after 100 chunks in Qwen2-57B-A14B-Instruct-IMat-GGUF/imatrix.dat [100]6.3803,[101]6.4046,[102]6.4739,[103]6.5511,[104]6.6015,[105]6.7051,[106]6.7713,[107]6.8045,[108]6.7913,[109]6.7797, save_imatrix: stored collected data after 110 chunks in Qwen2-57B-A14B-Instruct-IMat-GGUF/imatrix.dat [110]6.7746,[111]6.7326,[112]6.6627,[113]6.6131,[114]6.6655,[115]6.6699,[116]6.6776,[117]6.6962,[118]6.7280,[119]6.7310, save_imatrix: stored collected data after 120 chunks in Qwen2-57B-A14B-Instruct-IMat-GGUF/imatrix.dat [120]6.7268,[121]6.7397,[122]6.7134,[123]6.7513,[124]6.7813,[125]6.8029,[126]6.8390,[127]6.8695,[128]6.9092, save_imatrix: stored collected data after 128 chunks in Qwen2-57B-A14B-Instruct-IMat-GGUF/imatrix.dat llama_print_timings: load time = 7331.82 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 = 541654.75 ms / 65536 tokens ( 8.26 ms per token, 120.99 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 = 545922.50 ms / 65537 tokens Final estimate: PPL = 6.9092 +/- 0.09472