llama_model_loader: loaded meta data with 28 key-value pairs and 338 tensors from Qwen2-Math-1.5B-Instruct-IMat-GGUF/Qwen2-Math-1.5B-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 = qwen2 llama_model_loader: - kv 1: general.type str = model llama_model_loader: - kv 2: general.name str = Qwen2 Math 1.5B Instruct llama_model_loader: - kv 3: general.finetune str = Instruct llama_model_loader: - kv 4: general.basename str = Qwen2-Math llama_model_loader: - kv 5: general.size_label str = 1.5B llama_model_loader: - kv 6: general.license str = apache-2.0 llama_model_loader: - kv 7: general.tags arr[str,2] = ["chat", "text-generation"] llama_model_loader: - kv 8: general.languages arr[str,1] = ["en"] llama_model_loader: - kv 9: qwen2.block_count u32 = 28 llama_model_loader: - kv 10: qwen2.context_length u32 = 4096 llama_model_loader: - kv 11: qwen2.embedding_length u32 = 1536 llama_model_loader: - kv 12: qwen2.feed_forward_length u32 = 8960 llama_model_loader: - kv 13: qwen2.attention.head_count u32 = 12 llama_model_loader: - kv 14: qwen2.attention.head_count_kv u32 = 2 llama_model_loader: - kv 15: qwen2.rope.freq_base f32 = 10000.000000 llama_model_loader: - kv 16: qwen2.attention.layer_norm_rms_epsilon f32 = 0.000001 llama_model_loader: - kv 17: general.file_type u32 = 7 llama_model_loader: - kv 18: tokenizer.ggml.model str = gpt2 llama_model_loader: - kv 19: tokenizer.ggml.pre str = qwen2 llama_model_loader: - kv 20: tokenizer.ggml.tokens arr[str,151936] = ["!", "\"", "#", "$", "%", "&", "'", ... llama_model_loader: - kv 21: tokenizer.ggml.token_type arr[i32,151936] = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ... llama_model_loader: - kv 22: tokenizer.ggml.merges arr[str,151387] = ["Ġ Ġ", "ĠĠ ĠĠ", "i n", "Ġ t",... llama_model_loader: - kv 23: tokenizer.ggml.eos_token_id u32 = 151645 llama_model_loader: - kv 24: tokenizer.ggml.padding_token_id u32 = 151643 llama_model_loader: - kv 25: tokenizer.ggml.bos_token_id u32 = 151643 llama_model_loader: - kv 26: tokenizer.chat_template str = {% for message in messages %}{% if lo... llama_model_loader: - kv 27: general.quantization_version u32 = 2 llama_model_loader: - type f32: 141 tensors llama_model_loader: - type q8_0: 197 tensors llm_load_vocab: special tokens cache size = 3 llm_load_vocab: token to piece cache size = 0.9308 MB llm_load_print_meta: format = GGUF V3 (latest) llm_load_print_meta: arch = qwen2 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: vocab_only = 0 llm_load_print_meta: n_ctx_train = 4096 llm_load_print_meta: n_embd = 1536 llm_load_print_meta: n_layer = 28 llm_load_print_meta: n_head = 12 llm_load_print_meta: n_head_kv = 2 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 = 6 llm_load_print_meta: n_embd_k_gqa = 256 llm_load_print_meta: n_embd_v_gqa = 256 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 = 8960 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 = 2 llm_load_print_meta: rope scaling = linear llm_load_print_meta: freq_base_train = 10000.0 llm_load_print_meta: freq_scale_train = 1 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: model type = ?B llm_load_print_meta: model ftype = Q8_0 llm_load_print_meta: model params = 1.54 B llm_load_print_meta: model size = 1.53 GiB (8.50 BPW) llm_load_print_meta: general.name = Qwen2 Math 1.5B 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|>' 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 = 0.30 MiB llm_load_tensors: offloading 28 repeating layers to GPU llm_load_tensors: offloading non-repeating layers to GPU llm_load_tensors: offloaded 29/29 layers to GPU llm_load_tensors: CPU buffer size = 236.47 MiB llm_load_tensors: CUDA0 buffer size = 1564.63 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 = 1 llama_kv_cache_init: CUDA0 KV buffer size = 14.00 MiB llama_new_context_with_model: KV self size = 14.00 MiB, K (f16): 7.00 MiB, V (f16): 7.00 MiB llama_new_context_with_model: CUDA_Host output buffer size = 0.58 MiB llama_new_context_with_model: CUDA0 compute buffer size = 299.75 MiB llama_new_context_with_model: CUDA_Host compute buffer size = 4.01 MiB llama_new_context_with_model: graph nodes = 986 llama_new_context_with_model: graph splits = 2 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 129.312 ms compute_imatrix: computing over 128 chunks with batch_size 512 compute_imatrix: 0.41 seconds per pass - ETA 0.87 minutes [1]21.5954,[2]16.3481,[3]13.5135,[4]15.6791,[5]15.3211,[6]14.5537,[7]15.0464,[8]14.4367,[9]15.6930, save_imatrix: stored collected data after 10 chunks in Qwen2-Math-1.5B-Instruct-IMat-GGUF/imatrix.dat [10]14.8501,[11]14.1743,[12]15.7505,[13]17.8924,[14]18.3293,[15]20.2933,[16]20.9890,[17]21.6694,[18]23.3586,[19]22.5698, save_imatrix: stored collected data after 20 chunks in Qwen2-Math-1.5B-Instruct-IMat-GGUF/imatrix.dat [20]22.5694,[21]24.8524,[22]25.7592,[23]25.6477,[24]26.0525,[25]26.5780,[26]27.1280,[27]28.3387,[28]29.2826,[29]30.8846, save_imatrix: stored collected data after 30 chunks in Qwen2-Math-1.5B-Instruct-IMat-GGUF/imatrix.dat [30]31.7509,[31]32.9879,[32]32.0955,[33]31.0417,[34]30.2879,[35]29.3795,[36]32.0682,[37]36.4868,[38]37.8778,[39]38.0367, save_imatrix: stored collected data after 40 chunks in Qwen2-Math-1.5B-Instruct-IMat-GGUF/imatrix.dat [40]38.6583,[41]38.5933,[42]40.2867,[43]41.3568,[44]42.4316,[45]43.3600,[46]43.9424,[47]43.4594,[48]43.0973,[49]42.9915, save_imatrix: stored collected data after 50 chunks in Qwen2-Math-1.5B-Instruct-IMat-GGUF/imatrix.dat [50]42.5710,[51]42.0195,[52]41.9720,[53]43.2653,[54]43.5679,[55]44.4592,[56]44.7204,[57]44.5410,[58]44.5908,[59]44.4207, save_imatrix: stored collected data after 60 chunks in Qwen2-Math-1.5B-Instruct-IMat-GGUF/imatrix.dat [60]44.4606,[61]43.8949,[62]43.5317,[63]43.8146,[64]44.3940,[65]43.9416,[66]43.4245,[67]42.9840,[68]42.1103,[69]41.7746, save_imatrix: stored collected data after 70 chunks in Qwen2-Math-1.5B-Instruct-IMat-GGUF/imatrix.dat [70]41.2368,[71]40.5090,[72]39.9531,[73]39.5549,[74]38.6676,[75]37.9093,[76]37.1520,[77]36.5948,[78]36.2084,[79]35.8016, save_imatrix: stored collected data after 80 chunks in Qwen2-Math-1.5B-Instruct-IMat-GGUF/imatrix.dat [80]35.2852,[81]35.1617,[82]34.8628,[83]34.3655,[84]34.3487,[85]34.1451,[86]34.3092,[87]34.0145,[88]33.9010,[89]33.9726, save_imatrix: stored collected data after 90 chunks in Qwen2-Math-1.5B-Instruct-IMat-GGUF/imatrix.dat [90]34.0018,[91]33.9192,[92]33.1946,[93]32.6052,[94]31.9016,[95]31.2409,[96]30.6513,[97]30.0298,[98]29.4772,[99]29.6608, save_imatrix: stored collected data after 100 chunks in Qwen2-Math-1.5B-Instruct-IMat-GGUF/imatrix.dat [100]29.6859,[101]29.6662,[102]30.0276,[103]30.3328,[104]30.5797,[105]31.0761,[106]31.5270,[107]31.6387,[108]31.3431,[109]31.2995, save_imatrix: stored collected data after 110 chunks in Qwen2-Math-1.5B-Instruct-IMat-GGUF/imatrix.dat [110]31.2990,[111]30.8986,[112]30.5160,[113]30.5057,[114]30.5968,[115]30.7201,[116]30.7513,[117]30.8649,[118]30.9462,[119]30.8916, save_imatrix: stored collected data after 120 chunks in Qwen2-Math-1.5B-Instruct-IMat-GGUF/imatrix.dat [120]30.8107,[121]30.7677,[122]30.4821,[123]30.6267,[124]30.9271,[125]31.2403,[126]31.5599,[127]31.8815,[128]32.1625, save_imatrix: stored collected data after 128 chunks in Qwen2-Math-1.5B-Instruct-IMat-GGUF/imatrix.dat llama_print_timings: load time = 964.13 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 = 29971.58 ms / 65536 tokens ( 0.46 ms per token, 2186.60 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 = 31510.10 ms / 65537 tokens Final estimate: PPL = 32.1625 +/- 0.64699