File size: 12,359 Bytes
3f44c5b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
main: build = 2998 (9588f196)
main: built with cc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 for x86_64-linux-gnu
main: seed  = 1716654535
llama_model_loader: loaded meta data with 23 key-value pairs and 291 tensors from RoLlama2-7b-Chat-IMat-GGUF/RoLlama2-7b-Chat.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              = llama
llama_model_loader: - kv   1:                               general.name str              = RoLlama2-7b-Chat
llama_model_loader: - kv   2:                          llama.block_count u32              = 32
llama_model_loader: - kv   3:                       llama.context_length u32              = 4096
llama_model_loader: - kv   4:                     llama.embedding_length u32              = 4096
llama_model_loader: - kv   5:                  llama.feed_forward_length u32              = 11008
llama_model_loader: - kv   6:                 llama.attention.head_count u32              = 32
llama_model_loader: - kv   7:              llama.attention.head_count_kv u32              = 32
llama_model_loader: - kv   8:                       llama.rope.freq_base f32              = 10000.000000
llama_model_loader: - kv   9:     llama.attention.layer_norm_rms_epsilon f32              = 0.000010
llama_model_loader: - kv  10:                          general.file_type u32              = 0
llama_model_loader: - kv  11:                           llama.vocab_size u32              = 32004
llama_model_loader: - kv  12:                 llama.rope.dimension_count u32              = 128
llama_model_loader: - kv  13:                       tokenizer.ggml.model str              = llama
llama_model_loader: - kv  14:                         tokenizer.ggml.pre str              = default
llama_model_loader: - kv  15:                      tokenizer.ggml.tokens arr[str,32004]   = ["<unk>", "<s>", "</s>", "<0x00>", "<...
llama_model_loader: - kv  16:                      tokenizer.ggml.scores arr[f32,32004]   = [0.000000, 0.000000, 0.000000, 0.0000...
llama_model_loader: - kv  17:                  tokenizer.ggml.token_type arr[i32,32004]   = [2, 3, 3, 6, 6, 6, 6, 6, 6, 6, 6, 6, ...
llama_model_loader: - kv  18:                tokenizer.ggml.bos_token_id u32              = 1
llama_model_loader: - kv  19:                tokenizer.ggml.eos_token_id u32              = 2
llama_model_loader: - kv  20:               tokenizer.ggml.add_bos_token bool             = true
llama_model_loader: - kv  21:               tokenizer.ggml.add_eos_token bool             = false
llama_model_loader: - kv  22:               general.quantization_version u32              = 2
llama_model_loader: - type  f32:  291 tensors
llm_load_vocab: special tokens definition check successful ( 263/32004 ).
llm_load_print_meta: format           = GGUF V3 (latest)
llm_load_print_meta: arch             = llama
llm_load_print_meta: vocab type       = SPM
llm_load_print_meta: n_vocab          = 32004
llm_load_print_meta: n_merges         = 0
llm_load_print_meta: n_ctx_train      = 4096
llm_load_print_meta: n_embd           = 4096
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            = 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            = 1
llm_load_print_meta: n_embd_k_gqa     = 4096
llm_load_print_meta: n_embd_v_gqa     = 4096
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             = 11008
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  = 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       = 7B
llm_load_print_meta: model ftype      = all F32
llm_load_print_meta: model params     = 6.74 B
llm_load_print_meta: model size       = 25.10 GiB (32.00 BPW) 
llm_load_print_meta: general.name     = RoLlama2-7b-Chat
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: 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 20 repeating layers to GPU
llm_load_tensors: offloaded 20/33 layers to GPU
llm_load_tensors:        CPU buffer size = 25705.14 MiB
llm_load_tensors:      CUDA0 buffer size = 15440.62 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:  CUDA_Host KV buffer size =    96.00 MiB
llama_kv_cache_init:      CUDA0 KV buffer size =   160.00 MiB
llama_new_context_with_model: KV self size  =  256.00 MiB, K (f16):  128.00 MiB, V (f16):  128.00 MiB
llama_new_context_with_model:  CUDA_Host  output buffer size =     0.12 MiB
llama_new_context_with_model:      CUDA0 compute buffer size =   570.57 MiB
llama_new_context_with_model:  CUDA_Host compute buffer size =    17.01 MiB
llama_new_context_with_model: graph nodes  = 1030
llama_new_context_with_model: graph splits = 136

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 132.228 ms
compute_imatrix: computing over 234 chunks with batch_size 512
compute_imatrix: 1.05 seconds per pass - ETA 4.08 minutes
[1]5.9883,[2]4.3028,[3]4.2195,[4]4.7777,[5]5.3795,[6]5.4479,[7]4.9412,[8]5.3200,[9]5.5114,
save_imatrix: stored collected data after 10 chunks in RoLlama2-7b-Chat-IMat-GGUF/imatrix.dat
[10]5.8365,[11]5.8661,[12]5.3748,[13]5.5065,[14]5.4479,[15]5.8823,[16]6.0444,[17]6.3551,[18]6.5563,[19]6.7513,
save_imatrix: stored collected data after 20 chunks in RoLlama2-7b-Chat-IMat-GGUF/imatrix.dat
[20]6.9175,[21]6.9677,[22]7.0994,[23]6.7775,[24]6.5802,[25]6.6010,[26]6.3218,[27]6.1921,[28]5.9742,[29]5.9706,
save_imatrix: stored collected data after 30 chunks in RoLlama2-7b-Chat-IMat-GGUF/imatrix.dat
[30]6.0998,[31]6.1897,[32]6.2482,[33]6.2075,[34]6.2231,[35]6.2236,[36]5.9746,[37]5.8249,[38]5.7662,[39]5.7353,
save_imatrix: stored collected data after 40 chunks in RoLlama2-7b-Chat-IMat-GGUF/imatrix.dat
[40]5.6991,[41]5.6180,[42]5.6725,[43]5.6992,[44]5.7500,[45]5.8197,[46]5.8982,[47]5.9433,[48]6.0756,[49]6.1833,
save_imatrix: stored collected data after 50 chunks in RoLlama2-7b-Chat-IMat-GGUF/imatrix.dat
[50]6.3015,[51]6.3938,[52]6.5002,[53]6.4863,[54]6.4201,[55]6.3737,[56]6.4561,[57]6.5051,[58]6.5286,[59]6.5952,
save_imatrix: stored collected data after 60 chunks in RoLlama2-7b-Chat-IMat-GGUF/imatrix.dat
[60]6.6776,[61]6.7103,[62]6.7716,[63]6.8134,[64]6.8793,[65]6.8970,[66]6.9295,[67]6.9635,[68]6.9962,[69]7.0409,
save_imatrix: stored collected data after 70 chunks in RoLlama2-7b-Chat-IMat-GGUF/imatrix.dat
[70]7.0722,[71]7.1080,[72]7.1553,[73]7.1003,[74]7.0536,[75]7.0109,[76]6.9812,[77]6.9803,[78]6.9473,[79]6.9050,
save_imatrix: stored collected data after 80 chunks in RoLlama2-7b-Chat-IMat-GGUF/imatrix.dat
[80]6.8480,[81]6.8371,[82]6.7966,[83]6.7611,[84]6.7787,[85]6.8018,[86]6.8150,[87]6.8440,[88]6.8422,[89]6.8132,
save_imatrix: stored collected data after 90 chunks in RoLlama2-7b-Chat-IMat-GGUF/imatrix.dat
[90]6.8072,[91]6.8022,[92]6.7869,[93]6.7982,[94]6.7808,[95]6.7852,[96]6.8023,[97]6.8192,[98]6.7921,[99]6.7531,
save_imatrix: stored collected data after 100 chunks in RoLlama2-7b-Chat-IMat-GGUF/imatrix.dat
[100]6.7634,[101]6.7825,[102]6.7751,[103]6.7561,[104]6.7218,[105]6.7086,[106]6.7279,[107]6.7382,[108]6.7262,[109]6.7229,
save_imatrix: stored collected data after 110 chunks in RoLlama2-7b-Chat-IMat-GGUF/imatrix.dat
[110]6.7033,[111]6.7208,[112]6.7393,[113]6.7427,[114]6.7666,[115]6.7688,[116]6.7650,[117]6.7618,[118]6.7739,[119]6.7529,
save_imatrix: stored collected data after 120 chunks in RoLlama2-7b-Chat-IMat-GGUF/imatrix.dat
[120]6.7519,[121]6.7428,[122]6.7174,[123]6.7380,[124]6.7345,[125]6.7393,[126]6.7300,[127]6.7290,[128]6.7449,[129]6.7188,
save_imatrix: stored collected data after 130 chunks in RoLlama2-7b-Chat-IMat-GGUF/imatrix.dat
[130]6.6931,[131]6.6752,[132]6.6763,[133]6.6400,[134]6.6423,[135]6.6107,[136]6.5848,[137]6.5509,[138]6.5181,[139]6.4871,
save_imatrix: stored collected data after 140 chunks in RoLlama2-7b-Chat-IMat-GGUF/imatrix.dat
[140]6.4590,[141]6.4332,[142]6.4075,[143]6.4080,[144]6.4042,[145]6.3873,[146]6.3619,[147]6.3590,[148]6.3526,[149]6.3469,
save_imatrix: stored collected data after 150 chunks in RoLlama2-7b-Chat-IMat-GGUF/imatrix.dat
[150]6.3401,[151]6.3234,[152]6.3164,[153]6.3043,[154]6.2931,[155]6.3051,[156]6.2827,[157]6.2784,[158]6.2902,[159]6.2846,
save_imatrix: stored collected data after 160 chunks in RoLlama2-7b-Chat-IMat-GGUF/imatrix.dat
[160]6.2848,[161]6.2951,[162]6.2968,[163]6.3100,[164]6.3169,[165]6.3329,[166]6.3391,[167]6.3384,[168]6.3446,[169]6.3505,
save_imatrix: stored collected data after 170 chunks in RoLlama2-7b-Chat-IMat-GGUF/imatrix.dat
[170]6.3742,[171]6.3733,[172]6.3863,[173]6.4140,[174]6.4263,[175]6.4508,[176]6.4671,[177]6.4877,[178]6.5032,[179]6.5302,
save_imatrix: stored collected data after 180 chunks in RoLlama2-7b-Chat-IMat-GGUF/imatrix.dat
[180]6.5455,[181]6.5936,[182]6.6102,[183]6.6312,[184]6.6330,[185]6.6379,[186]6.6479,[187]6.6518,[188]6.6372,[189]6.6427,
save_imatrix: stored collected data after 190 chunks in RoLlama2-7b-Chat-IMat-GGUF/imatrix.dat
[190]6.6507,[191]6.6596,[192]6.6670,[193]6.6965,[194]6.6913,[195]6.6699,[196]6.7046,[197]6.7383,[198]6.7669,[199]6.8205,
save_imatrix: stored collected data after 200 chunks in RoLlama2-7b-Chat-IMat-GGUF/imatrix.dat
[200]6.8660,[201]6.8749,[202]6.8785,[203]6.8443,[204]6.8471,[205]6.8582,[206]6.8843,[207]6.8737,[208]6.8706,[209]6.8729,
save_imatrix: stored collected data after 210 chunks in RoLlama2-7b-Chat-IMat-GGUF/imatrix.dat
[210]6.8845,[211]6.9005,[212]6.8986,[213]6.8943,[214]6.8991,[215]6.9194,[216]6.9322,[217]6.9379,[218]6.9316,[219]6.9273,
save_imatrix: stored collected data after 220 chunks in RoLlama2-7b-Chat-IMat-GGUF/imatrix.dat
[220]6.9200,[221]6.9196,[222]6.9186,[223]6.9421,[224]6.9281,[225]6.9293,[226]6.9200,[227]6.9536,[228]6.9927,[229]7.0349,
save_imatrix: stored collected data after 230 chunks in RoLlama2-7b-Chat-IMat-GGUF/imatrix.dat
[230]7.0738,[231]7.0879,[232]7.0712,[233]7.0533,[234]7.0337,
save_imatrix: stored collected data after 234 chunks in RoLlama2-7b-Chat-IMat-GGUF/imatrix.dat

llama_print_timings:        load time =    2820.15 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 =  232394.01 ms / 119808 tokens (    1.94 ms per token,   515.54 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 =  235165.91 ms / 119809 tokens

Final estimate: PPL = 7.0337 +/- 0.06898