File size: 12,057 Bytes
a30c7d6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
main: build = 3008 (1d8fca72)
main: built with cc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 for x86_64-linux-gnu
main: seed  = 1716818940
llama_model_loader: loaded meta data with 22 key-value pairs and 291 tensors from internlm2-math-plus-7b-IMat-GGUF/internlm2-math-plus-7b.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              = internlm2
llama_model_loader: - kv   1:                               general.name str              = InternLM2
llama_model_loader: - kv   2:                   internlm2.context_length u32              = 8192
llama_model_loader: - kv   3:                      internlm2.block_count u32              = 32
llama_model_loader: - kv   4:                 internlm2.embedding_length u32              = 4096
llama_model_loader: - kv   5:              internlm2.feed_forward_length u32              = 14336
llama_model_loader: - kv   6:                   internlm2.rope.freq_base f32              = 1000000.000000
llama_model_loader: - kv   7:             internlm2.attention.head_count u32              = 32
llama_model_loader: - kv   8: internlm2.attention.layer_norm_rms_epsilon f32              = 0.000010
llama_model_loader: - kv   9:          internlm2.attention.head_count_kv u32              = 8
llama_model_loader: - kv  10:                          general.file_type u32              = 0
llama_model_loader: - kv  11:                       tokenizer.ggml.model str              = llama
llama_model_loader: - kv  12:                         tokenizer.ggml.pre str              = default
llama_model_loader: - kv  13:                      tokenizer.ggml.tokens arr[str,92544]   = ["<unk>", "<s>", "</s>", "<0x00>", "<...
llama_model_loader: - kv  14:                      tokenizer.ggml.scores arr[f32,92544]   = [0.000000, 0.000000, 0.000000, 0.0000...
llama_model_loader: - kv  15:                  tokenizer.ggml.token_type arr[i32,92544]   = [2, 3, 3, 6, 6, 6, 6, 6, 6, 6, 6, 6, ...
llama_model_loader: - kv  16:            tokenizer.ggml.add_space_prefix bool             = false
llama_model_loader: - kv  17:                tokenizer.ggml.bos_token_id u32              = 1
llama_model_loader: - kv  18:                tokenizer.ggml.eos_token_id u32              = 2
llama_model_loader: - kv  19:            tokenizer.ggml.padding_token_id u32              = 2
llama_model_loader: - kv  20:                    tokenizer.chat_template str              = {{ bos_token }}{% for message in mess...
llama_model_loader: - kv  21:               general.quantization_version u32              = 2
llama_model_loader: - type  f32:  291 tensors
llm_load_vocab: mismatch in special tokens definition ( 405/92544 vs 259/92544 ).
llm_load_print_meta: format           = GGUF V3 (latest)
llm_load_print_meta: arch             = internlm2
llm_load_print_meta: vocab type       = SPM
llm_load_print_meta: n_vocab          = 92544
llm_load_print_meta: n_merges         = 0
llm_load_print_meta: n_ctx_train      = 8192
llm_load_print_meta: n_embd           = 4096
llm_load_print_meta: n_head           = 32
llm_load_print_meta: n_head_kv        = 8
llm_load_print_meta: n_layer          = 32
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            = 4
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             = 14336
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  = 1000000.0
llm_load_print_meta: freq_scale_train = 1
llm_load_print_meta: n_yarn_orig_ctx  = 8192
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       = 7B
llm_load_print_meta: model ftype      = all F32
llm_load_print_meta: model params     = 7.74 B
llm_load_print_meta: model size       = 28.83 GiB (32.00 BPW) 
llm_load_print_meta: general.name     = InternLM2
llm_load_print_meta: BOS token        = 1 '<s>'
llm_load_print_meta: EOS token        = 2 '</s>'
llm_load_print_meta: UNK token        = 0 '<unk>'
llm_load_print_meta: PAD token        = 2 '</s>'
llm_load_print_meta: LF token         = 13 '<0x0A>'
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 =    0.30 MiB
llm_load_tensors: offloading 24 repeating layers to GPU
llm_load_tensors: offloaded 24/33 layers to GPU
llm_load_tensors:        CPU buffer size = 29517.02 MiB
llm_load_tensors:      CUDA0 buffer size = 19968.75 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 =    16.00 MiB
llama_kv_cache_init:      CUDA0 KV buffer size =    48.00 MiB
llama_new_context_with_model: KV self size  =   64.00 MiB, K (f16):   32.00 MiB, V (f16):   32.00 MiB
llama_new_context_with_model:  CUDA_Host  output buffer size =     0.35 MiB
llama_new_context_with_model:      CUDA0 compute buffer size =  1634.75 MiB
llama_new_context_with_model:  CUDA_Host compute buffer size =     9.01 MiB
llama_new_context_with_model: graph nodes  = 1030
llama_new_context_with_model: graph splits = 92

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 145.596 ms
compute_imatrix: computing over 209 chunks with batch_size 512
compute_imatrix: 0.96 seconds per pass - ETA 3.32 minutes
[1]8.6168,[2]6.0773,[3]5.4163,[4]6.1982,[5]6.3812,[6]5.6271,[7]6.3406,[8]6.3271,[9]7.0202,
save_imatrix: stored collected data after 10 chunks in internlm2-math-plus-7b-IMat-GGUF/imatrix.dat
[10]7.2999,[11]7.9082,[12]8.0650,[13]9.5172,[14]9.9489,[15]10.7399,[16]11.0800,[17]11.6268,[18]11.0305,[19]11.5148,
save_imatrix: stored collected data after 20 chunks in internlm2-math-plus-7b-IMat-GGUF/imatrix.dat
[20]11.4900,[21]10.8624,[22]10.9906,[23]10.2198,[24]9.8705,[25]9.3718,[26]9.2935,[27]9.7640,[28]9.7553,[29]10.1653,
save_imatrix: stored collected data after 30 chunks in internlm2-math-plus-7b-IMat-GGUF/imatrix.dat
[30]10.4245,[31]10.3090,[32]9.7674,[33]9.4135,[34]9.3314,[35]9.2103,[36]9.0950,[37]9.3619,[38]9.6158,[39]9.8277,
save_imatrix: stored collected data after 40 chunks in internlm2-math-plus-7b-IMat-GGUF/imatrix.dat
[40]10.1027,[41]10.3507,[42]10.6739,[43]10.9750,[44]11.2557,[45]11.3919,[46]11.4866,[47]11.4222,[48]11.3479,[49]11.5629,
save_imatrix: stored collected data after 50 chunks in internlm2-math-plus-7b-IMat-GGUF/imatrix.dat
[50]11.7936,[51]11.8713,[52]12.1181,[53]12.1751,[54]12.1674,[55]12.0785,[56]12.0195,[57]11.9204,[58]11.9528,[59]11.8437,
save_imatrix: stored collected data after 60 chunks in internlm2-math-plus-7b-IMat-GGUF/imatrix.dat
[60]11.8499,[61]12.0561,[62]12.2897,[63]12.6332,[64]12.6103,[65]12.4497,[66]12.1780,[67]11.9540,[68]11.7427,[69]11.5259,
save_imatrix: stored collected data after 70 chunks in internlm2-math-plus-7b-IMat-GGUF/imatrix.dat
[70]11.4201,[71]11.1605,[72]10.9335,[73]10.6764,[74]10.8084,[75]10.9271,[76]10.9531,[77]10.9341,[78]11.0427,[79]11.0051,
save_imatrix: stored collected data after 80 chunks in internlm2-math-plus-7b-IMat-GGUF/imatrix.dat
[80]11.0146,[81]10.9573,[82]10.9558,[83]10.9465,[84]10.9364,[85]10.9364,[86]10.9115,[87]10.9393,[88]10.9545,[89]11.0222,
save_imatrix: stored collected data after 90 chunks in internlm2-math-plus-7b-IMat-GGUF/imatrix.dat
[90]11.0829,[91]11.1312,[92]11.0959,[93]11.0311,[94]10.9948,[95]10.9958,[96]10.9463,[97]10.9640,[98]10.9649,[99]10.9665,
save_imatrix: stored collected data after 100 chunks in internlm2-math-plus-7b-IMat-GGUF/imatrix.dat
[100]10.9076,[101]10.9234,[102]10.8731,[103]10.7965,[104]10.7553,[105]10.7489,[106]10.6998,[107]10.6522,[108]10.6281,[109]10.6360,
save_imatrix: stored collected data after 110 chunks in internlm2-math-plus-7b-IMat-GGUF/imatrix.dat
[110]10.6586,[111]10.6372,[112]10.6652,[113]10.6926,[114]10.6349,[115]10.5984,[116]10.6427,[117]10.6177,[118]10.6488,[119]10.5550,
save_imatrix: stored collected data after 120 chunks in internlm2-math-plus-7b-IMat-GGUF/imatrix.dat
[120]10.4760,[121]10.3803,[122]10.2836,[123]10.1938,[124]10.1074,[125]10.0241,[126]9.9999,[127]9.9689,[128]9.9235,[129]9.8538,
save_imatrix: stored collected data after 130 chunks in internlm2-math-plus-7b-IMat-GGUF/imatrix.dat
[130]9.8316,[131]9.7957,[132]9.7496,[133]9.7198,[134]9.6710,[135]9.6351,[136]9.6114,[137]9.5686,[138]9.5217,[139]9.5032,
save_imatrix: stored collected data after 140 chunks in internlm2-math-plus-7b-IMat-GGUF/imatrix.dat
[140]9.4517,[141]9.4110,[142]9.5089,[143]9.6658,[144]9.8620,[145]10.0062,[146]10.0122,[147]10.0509,[148]10.0944,[149]10.1520,
save_imatrix: stored collected data after 150 chunks in internlm2-math-plus-7b-IMat-GGUF/imatrix.dat
[150]10.1926,[151]10.2064,[152]10.2056,[153]10.2728,[154]10.3324,[155]10.4128,[156]10.4584,[157]10.5164,[158]10.5840,[159]10.6087,
save_imatrix: stored collected data after 160 chunks in internlm2-math-plus-7b-IMat-GGUF/imatrix.dat
[160]10.6699,[161]10.7003,[162]10.7507,[163]10.7984,[164]10.8557,[165]10.8774,[166]10.9137,[167]10.9503,[168]10.9603,[169]10.9802,
save_imatrix: stored collected data after 170 chunks in internlm2-math-plus-7b-IMat-GGUF/imatrix.dat
[170]11.0021,[171]11.0323,[172]11.0788,[173]11.1006,[174]11.0747,[175]11.1685,[176]11.2557,[177]11.3494,[178]11.4653,[179]11.5544,
save_imatrix: stored collected data after 180 chunks in internlm2-math-plus-7b-IMat-GGUF/imatrix.dat
[180]11.5983,[181]11.5868,[182]11.6085,[183]11.6615,[184]11.7197,[185]11.7194,[186]11.7281,[187]11.7453,[188]11.7771,[189]11.7798,
save_imatrix: stored collected data after 190 chunks in internlm2-math-plus-7b-IMat-GGUF/imatrix.dat
[190]11.7851,[191]11.8087,[192]11.8173,[193]11.8368,[194]11.8177,[195]11.8339,[196]11.8145,[197]11.8451,[198]11.8571,[199]11.9365,
save_imatrix: stored collected data after 200 chunks in internlm2-math-plus-7b-IMat-GGUF/imatrix.dat
[200]11.8686,[201]11.9083,[202]11.8877,[203]11.9848,[204]12.0625,[205]12.1544,[206]12.2099,[207]12.2741,[208]12.2187,[209]12.1662,
save_imatrix: stored collected data after 209 chunks in internlm2-math-plus-7b-IMat-GGUF/imatrix.dat

llama_print_timings:        load time =    2815.28 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 =  184148.18 ms / 107008 tokens (    1.72 ms per token,   581.10 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 =  187280.86 ms / 107009 tokens

Final estimate: PPL = 12.1662 +/- 0.16829