File size: 13,076 Bytes
342327d |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 |
main: build = 3003 (d298382a)
main: built with cc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 for x86_64-linux-gnu
main: seed = 1716752123
llama_model_loader: loaded meta data with 27 key-value pairs and 197 tensors from Phi-3-mini-128k-instruct-IMat-GGUF/Phi-3-mini-128k-instruct.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 = phi3
llama_model_loader: - kv 1: general.name str = Phi3
llama_model_loader: - kv 2: phi3.context_length u32 = 131072
llama_model_loader: - kv 3: phi3.rope.scaling.original_context_length u32 = 4096
llama_model_loader: - kv 4: phi3.embedding_length u32 = 3072
llama_model_loader: - kv 5: phi3.feed_forward_length u32 = 8192
llama_model_loader: - kv 6: phi3.block_count u32 = 32
llama_model_loader: - kv 7: phi3.attention.head_count u32 = 32
llama_model_loader: - kv 8: phi3.attention.head_count_kv u32 = 32
llama_model_loader: - kv 9: phi3.attention.layer_norm_rms_epsilon f32 = 0.000010
llama_model_loader: - kv 10: phi3.rope.dimension_count u32 = 96
llama_model_loader: - kv 11: phi3.rope.freq_base f32 = 10000.000000
llama_model_loader: - kv 12: general.file_type u32 = 0
llama_model_loader: - kv 13: phi3.rope.scaling.attn_factor f32 = 1.190238
llama_model_loader: - kv 14: tokenizer.ggml.model str = llama
llama_model_loader: - kv 15: tokenizer.ggml.pre str = default
llama_model_loader: - kv 16: tokenizer.ggml.tokens arr[str,32064] = ["<unk>", "<s>", "</s>", "<0x00>", "<...
llama_model_loader: - kv 17: tokenizer.ggml.scores arr[f32,32064] = [-1000.000000, -1000.000000, -1000.00...
llama_model_loader: - kv 18: tokenizer.ggml.token_type arr[i32,32064] = [3, 3, 4, 6, 6, 6, 6, 6, 6, 6, 6, 6, ...
llama_model_loader: - kv 19: tokenizer.ggml.bos_token_id u32 = 1
llama_model_loader: - kv 20: tokenizer.ggml.eos_token_id u32 = 32000
llama_model_loader: - kv 21: tokenizer.ggml.unknown_token_id u32 = 0
llama_model_loader: - kv 22: tokenizer.ggml.padding_token_id u32 = 32000
llama_model_loader: - kv 23: tokenizer.ggml.add_bos_token bool = true
llama_model_loader: - kv 24: tokenizer.ggml.add_eos_token bool = false
llama_model_loader: - kv 25: tokenizer.chat_template str = {{ bos_token }}{% for message in mess...
llama_model_loader: - kv 26: general.quantization_version u32 = 2
llama_model_loader: - type f32: 197 tensors
llm_load_vocab: special tokens definition check successful ( 323/32064 ).
llm_load_print_meta: format = GGUF V3 (latest)
llm_load_print_meta: arch = phi3
llm_load_print_meta: vocab type = SPM
llm_load_print_meta: n_vocab = 32064
llm_load_print_meta: n_merges = 0
llm_load_print_meta: n_ctx_train = 131072
llm_load_print_meta: n_embd = 3072
llm_load_print_meta: n_head = 32
llm_load_print_meta: n_head_kv = 32
llm_load_print_meta: n_layer = 32
llm_load_print_meta: n_rot = 96
llm_load_print_meta: n_embd_head_k = 96
llm_load_print_meta: n_embd_head_v = 96
llm_load_print_meta: n_gqa = 1
llm_load_print_meta: n_embd_k_gqa = 3072
llm_load_print_meta: n_embd_v_gqa = 3072
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 = 8192
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_yarn_orig_ctx = 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 = 3B
llm_load_print_meta: model ftype = all F32
llm_load_print_meta: model params = 3.82 B
llm_load_print_meta: model size = 14.23 GiB (32.00 BPW)
llm_load_print_meta: general.name = Phi3
llm_load_print_meta: BOS token = 1 '<s>'
llm_load_print_meta: EOS token = 32000 '<|endoftext|>'
llm_load_print_meta: UNK token = 0 '<unk>'
llm_load_print_meta: PAD token = 32000 '<|endoftext|>'
llm_load_print_meta: LF token = 13 '<0x0A>'
llm_load_print_meta: EOT token = 32007 '<|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 = 0.22 MiB
llm_load_tensors: offloading 32 repeating layers to GPU
llm_load_tensors: offloading non-repeating layers to GPU
llm_load_tensors: offloaded 33/33 layers to GPU
llm_load_tensors: CPU buffer size = 375.75 MiB
llm_load_tensors: CUDA0 buffer size = 14200.53 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 = 192.00 MiB
llama_new_context_with_model: KV self size = 192.00 MiB, K (f16): 96.00 MiB, V (f16): 96.00 MiB
llama_new_context_with_model: CUDA_Host output buffer size = 0.12 MiB
llama_new_context_with_model: CUDA0 compute buffer size = 83.00 MiB
llama_new_context_with_model: CUDA_Host compute buffer size = 7.01 MiB
llama_new_context_with_model: graph nodes = 1286
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 133.64 ms
compute_imatrix: computing over 234 chunks with batch_size 512
compute_imatrix: 0.32 seconds per pass - ETA 1.23 minutes
[1]6.0727,[2]4.4610,[3]4.4629,[4]4.9370,[5]5.3244,[6]5.4170,[7]4.8496,[8]5.2827,[9]5.5966,
save_imatrix: stored collected data after 10 chunks in Phi-3-mini-128k-instruct-IMat-GGUF/imatrix.dat
[10]5.9047,[11]5.8828,[12]5.4226,[13]5.5632,[14]5.4485,[15]5.8942,[16]5.9986,[17]6.2966,[18]6.4616,[19]6.6562,
save_imatrix: stored collected data after 20 chunks in Phi-3-mini-128k-instruct-IMat-GGUF/imatrix.dat
[20]6.8072,[21]6.8799,[22]7.1044,[23]6.8300,[24]6.6506,[25]6.6546,[26]6.2990,[27]6.0381,[28]5.7430,[29]5.7002,
save_imatrix: stored collected data after 30 chunks in Phi-3-mini-128k-instruct-IMat-GGUF/imatrix.dat
[30]5.8055,[31]5.8773,[32]5.9258,[33]5.9168,[34]5.9704,[35]5.9718,[36]5.7531,[37]5.6171,[38]5.5492,[39]5.5208,
save_imatrix: stored collected data after 40 chunks in Phi-3-mini-128k-instruct-IMat-GGUF/imatrix.dat
[40]5.4923,[41]5.4249,[42]5.4651,[43]5.5062,[44]5.5516,[45]5.6231,[46]5.7071,[47]5.7971,[48]5.9323,[49]6.0424,
save_imatrix: stored collected data after 50 chunks in Phi-3-mini-128k-instruct-IMat-GGUF/imatrix.dat
[50]6.1599,[51]6.2644,[52]6.3634,[53]6.3316,[54]6.2462,[55]6.1791,[56]6.2703,[57]6.3211,[58]6.3341,[59]6.3940,
save_imatrix: stored collected data after 60 chunks in Phi-3-mini-128k-instruct-IMat-GGUF/imatrix.dat
[60]6.4760,[61]6.5040,[62]6.5788,[63]6.6285,[64]6.7074,[65]6.7470,[66]6.7897,[67]6.8378,[68]6.8799,[69]6.9477,
save_imatrix: stored collected data after 70 chunks in Phi-3-mini-128k-instruct-IMat-GGUF/imatrix.dat
[70]6.9901,[71]7.0393,[72]7.0741,[73]7.0324,[74]6.9788,[75]6.9180,[76]6.8588,[77]6.8484,[78]6.7958,[79]6.7419,
save_imatrix: stored collected data after 80 chunks in Phi-3-mini-128k-instruct-IMat-GGUF/imatrix.dat
[80]6.6785,[81]6.6562,[82]6.6094,[83]6.5719,[84]6.5868,[85]6.6086,[86]6.6214,[87]6.6579,[88]6.6725,[89]6.6531,
save_imatrix: stored collected data after 90 chunks in Phi-3-mini-128k-instruct-IMat-GGUF/imatrix.dat
[90]6.6234,[91]6.6468,[92]6.6571,[93]6.6760,[94]6.6899,[95]6.7034,[96]6.7345,[97]6.7548,[98]6.7283,[99]6.6884,
save_imatrix: stored collected data after 100 chunks in Phi-3-mini-128k-instruct-IMat-GGUF/imatrix.dat
[100]6.7032,[101]6.7244,[102]6.7139,[103]6.6792,[104]6.6236,[105]6.6084,[106]6.6134,[107]6.6201,[108]6.5983,[109]6.5869,
save_imatrix: stored collected data after 110 chunks in Phi-3-mini-128k-instruct-IMat-GGUF/imatrix.dat
[110]6.5668,[111]6.5736,[112]6.5833,[113]6.5814,[114]6.5906,[115]6.5858,[116]6.5842,[117]6.5771,[118]6.5829,[119]6.5621,
save_imatrix: stored collected data after 120 chunks in Phi-3-mini-128k-instruct-IMat-GGUF/imatrix.dat
[120]6.5647,[121]6.5506,[122]6.5262,[123]6.5435,[124]6.5348,[125]6.5383,[126]6.5255,[127]6.5257,[128]6.5349,[129]6.5169,
save_imatrix: stored collected data after 130 chunks in Phi-3-mini-128k-instruct-IMat-GGUF/imatrix.dat
[130]6.4927,[131]6.4831,[132]6.4802,[133]6.4305,[134]6.4382,[135]6.4156,[136]6.3964,[137]6.3724,[138]6.3466,[139]6.3174,
save_imatrix: stored collected data after 140 chunks in Phi-3-mini-128k-instruct-IMat-GGUF/imatrix.dat
[140]6.2951,[141]6.2763,[142]6.2547,[143]6.2554,[144]6.2527,[145]6.2348,[146]6.2123,[147]6.2092,[148]6.1993,[149]6.1886,
save_imatrix: stored collected data after 150 chunks in Phi-3-mini-128k-instruct-IMat-GGUF/imatrix.dat
[150]6.1821,[151]6.1677,[152]6.1637,[153]6.1538,[154]6.1412,[155]6.1651,[156]6.1401,[157]6.1342,[158]6.1511,[159]6.1461,
save_imatrix: stored collected data after 160 chunks in Phi-3-mini-128k-instruct-IMat-GGUF/imatrix.dat
[160]6.1502,[161]6.1632,[162]6.1662,[163]6.1865,[164]6.1987,[165]6.2188,[166]6.2278,[167]6.2257,[168]6.2270,[169]6.2340,
save_imatrix: stored collected data after 170 chunks in Phi-3-mini-128k-instruct-IMat-GGUF/imatrix.dat
[170]6.2467,[171]6.2367,[172]6.2370,[173]6.2540,[174]6.2566,[175]6.2740,[176]6.2834,[177]6.2939,[178]6.3006,[179]6.3317,
save_imatrix: stored collected data after 180 chunks in Phi-3-mini-128k-instruct-IMat-GGUF/imatrix.dat
[180]6.3424,[181]6.3923,[182]6.4110,[183]6.4381,[184]6.4432,[185]6.4486,[186]6.4541,[187]6.4579,[188]6.4483,[189]6.4521,
save_imatrix: stored collected data after 190 chunks in Phi-3-mini-128k-instruct-IMat-GGUF/imatrix.dat
[190]6.4586,[191]6.4720,[192]6.4769,[193]6.5062,[194]6.4950,[195]6.4664,[196]6.5076,[197]6.5460,[198]6.5764,[199]6.6271,
save_imatrix: stored collected data after 200 chunks in Phi-3-mini-128k-instruct-IMat-GGUF/imatrix.dat
[200]6.6740,[201]6.6813,[202]6.6861,[203]6.6440,[204]6.6408,[205]6.6470,[206]6.6681,[207]6.6649,[208]6.6679,[209]6.6688,
save_imatrix: stored collected data after 210 chunks in Phi-3-mini-128k-instruct-IMat-GGUF/imatrix.dat
[210]6.6772,[211]6.6909,[212]6.6907,[213]6.6877,[214]6.6944,[215]6.7130,[216]6.7305,[217]6.7338,[218]6.7346,[219]6.7292,
save_imatrix: stored collected data after 220 chunks in Phi-3-mini-128k-instruct-IMat-GGUF/imatrix.dat
[220]6.7205,[221]6.7193,[222]6.7178,[223]6.7324,[224]6.7153,[225]6.7209,[226]6.7047,[227]6.7403,[228]6.7806,[229]6.8252,
save_imatrix: stored collected data after 230 chunks in Phi-3-mini-128k-instruct-IMat-GGUF/imatrix.dat
[230]6.8657,[231]6.8871,[232]6.8663,[233]6.8447,[234]6.8193,
save_imatrix: stored collected data after 234 chunks in Phi-3-mini-128k-instruct-IMat-GGUF/imatrix.dat
llama_print_timings: load time = 2127.07 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 = 52676.50 ms / 119808 tokens ( 0.44 ms per token, 2274.41 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 = 55109.70 ms / 119809 tokens
Final estimate: PPL = 6.8193 +/- 0.07007
|