Hugging Face's logo

legraphista
/
gemma-2-27b-it-IMat-GGUF

Text Generation
GGUF
quantized
GGUF
quantization
imat
imatrix
static
16bit
8bit
6bit
5bit
4bit
3bit
2bit
1bit
conversational
Model card Files Community
2
legraphista commited on
Commit
4582dc8
·
verified ·
1 Parent(s): 80a5944

Upload imatrix.log with huggingface_hub

Browse files
Files changed (1) hide show
  1. imatrix.log +58 -70
imatrix.log CHANGED
@@ -1,48 +1,62 @@
1
- llama_model_loader: loaded meta data with 28 key-value pairs and 508 tensors from gemma-2-27b-it-IMat-GGUF/gemma-2-27b-it.Q8_0.gguf.hardlink.gguf (version GGUF V3 (latest))
 
2
  llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output.
3
  llama_model_loader: - kv 0: general.architecture str = gemma2
4
- llama_model_loader: - kv 1: general.name str = gemma-2-27b-it
5
- llama_model_loader: - kv 2: gemma2.context_length u32 = 8192
6
- llama_model_loader: - kv 3: gemma2.embedding_length u32 = 4608
7
- llama_model_loader: - kv 4: gemma2.block_count u32 = 46
8
- llama_model_loader: - kv 5: gemma2.feed_forward_length u32 = 36864
9
- llama_model_loader: - kv 6: gemma2.attention.head_count u32 = 32
10
- llama_model_loader: - kv 7: gemma2.attention.head_count_kv u32 = 16
11
- llama_model_loader: - kv 8: gemma2.attention.layer_norm_rms_epsilon f32 = 0.000001
12
- llama_model_loader: - kv 9: gemma2.attention.key_length u32 = 128
13
- llama_model_loader: - kv 10: gemma2.attention.value_length u32 = 128
14
- llama_model_loader: - kv 11: general.file_type u32 = 7
15
- llama_model_loader: - kv 12: gemma2.attn_logit_softcapping f32 = 50.000000
16
- llama_model_loader: - kv 13: gemma2.final_logit_softcapping f32 = 30.000000
17
- llama_model_loader: - kv 14: tokenizer.ggml.model str = llama
18
- llama_model_loader: - kv 15: tokenizer.ggml.pre str = default
19
- llama_model_loader: - kv 16: tokenizer.ggml.tokens arr[str,256000] = ["<pad>", "<eos>", "<bos>", "<unk>", ...
20
- llama_model_loader: - kv 17: tokenizer.ggml.scores arr[f32,256000] = [-1000.000000, -1000.000000, -1000.00...
21
- llama_model_loader: - kv 18: tokenizer.ggml.token_type arr[i32,256000] = [3, 3, 3, 3, 1, 1, 1, 1, 1, 1, 1, 1, ...
22
- llama_model_loader: - kv 19: tokenizer.ggml.bos_token_id u32 = 2
23
- llama_model_loader: - kv 20: tokenizer.ggml.eos_token_id u32 = 1
24
- llama_model_loader: - kv 21: tokenizer.ggml.unknown_token_id u32 = 3
25
- llama_model_loader: - kv 22: tokenizer.ggml.padding_token_id u32 = 0
26
- llama_model_loader: - kv 23: tokenizer.ggml.add_bos_token bool = true
27
- llama_model_loader: - kv 24: tokenizer.ggml.add_eos_token bool = false
28
- llama_model_loader: - kv 25: tokenizer.chat_template str = {{ bos_token }}{% if messages[0]['rol...
29
- llama_model_loader: - kv 26: tokenizer.ggml.add_space_prefix bool = false
30
- llama_model_loader: - kv 27: general.quantization_version u32 = 2
 
 
 
 
 
 
 
 
 
 
 
31
  llama_model_loader: - type f32: 185 tensors
32
  llama_model_loader: - type q8_0: 323 tensors
33
- llm_load_vocab: special tokens cache size = 261
34
  llm_load_vocab: token to piece cache size = 1.6014 MB
35
  llm_load_print_meta: format = GGUF V3 (latest)
36
  llm_load_print_meta: arch = gemma2
37
  llm_load_print_meta: vocab type = SPM
38
  llm_load_print_meta: n_vocab = 256000
39
  llm_load_print_meta: n_merges = 0
 
40
  llm_load_print_meta: n_ctx_train = 8192
41
  llm_load_print_meta: n_embd = 4608
 
42
  llm_load_print_meta: n_head = 32
43
  llm_load_print_meta: n_head_kv = 16
44
- llm_load_print_meta: n_layer = 46
45
- llm_load_print_meta: n_rot = 144
46
  llm_load_print_meta: n_embd_head_k = 128
47
  llm_load_print_meta: n_embd_head_v = 128
48
  llm_load_print_meta: n_gqa = 2
@@ -68,18 +82,19 @@ llm_load_print_meta: ssm_d_conv = 0
68
  llm_load_print_meta: ssm_d_inner = 0
69
  llm_load_print_meta: ssm_d_state = 0
70
  llm_load_print_meta: ssm_dt_rank = 0
 
71
  llm_load_print_meta: model type = 27B
72
  llm_load_print_meta: model ftype = Q8_0
73
  llm_load_print_meta: model params = 27.23 B
74
  llm_load_print_meta: model size = 26.94 GiB (8.50 BPW)
75
- llm_load_print_meta: general.name = gemma-2-27b-it
76
  llm_load_print_meta: BOS token = 2 '<bos>'
77
  llm_load_print_meta: EOS token = 1 '<eos>'
78
  llm_load_print_meta: UNK token = 3 '<unk>'
79
  llm_load_print_meta: PAD token = 0 '<pad>'
80
  llm_load_print_meta: LF token = 227 '<0x0A>'
81
  llm_load_print_meta: EOT token = 107 '<end_of_turn>'
82
- llm_load_print_meta: max token length = 93
83
  ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no
84
  ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
85
  ggml_cuda_init: found 1 CUDA devices:
@@ -101,46 +116,19 @@ llama_kv_cache_init: CUDA0 KV buffer size = 148.00 MiB
101
  llama_new_context_with_model: KV self size = 184.00 MiB, K (f16): 92.00 MiB, V (f16): 92.00 MiB
102
  llama_new_context_with_model: CUDA_Host output buffer size = 0.98 MiB
103
  llama_new_context_with_model: CUDA0 compute buffer size = 1704.31 MiB
104
- llama_new_context_with_model: CUDA_Host compute buffer size = 10.01 MiB
105
  llama_new_context_with_model: graph nodes = 1850
106
  llama_new_context_with_model: graph splits = 121
107
 
108
- 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 |
109
  compute_imatrix: tokenizing the input ..
110
- compute_imatrix: tokenization took 94.256 ms
111
  compute_imatrix: computing over 128 chunks with batch_size 512
112
- compute_imatrix: 1.92 seconds per pass - ETA 4.08 minutes
113
- [1]12.2429,[2]6.2081,[3]5.2588,[4]6.2085,[5]6.7166,[6]7.2390,[7]7.6746,[8]8.1610,[9]8.5380,
114
- save_imatrix: stored collected data after 10 chunks in gemma-2-27b-it-IMat-GGUF/imatrix.dat
115
- [10]7.7409,[11]7.6231,[12]8.3069,[13]8.8175,[14]8.9820,[15]9.5963,[16]9.7448,[17]9.8649,[18]10.2368,[19]10.1102,
116
- save_imatrix: stored collected data after 20 chunks in gemma-2-27b-it-IMat-GGUF/imatrix.dat
117
- [20]10.2363,[21]10.9791,[22]10.9424,[23]10.8230,[24]11.0184,[25]10.9711,[26]10.8238,[27]11.0156,[28]11.1952,[29]11.2073,
118
- save_imatrix: stored collected data after 30 chunks in gemma-2-27b-it-IMat-GGUF/imatrix.dat
119
- [30]11.4849,[31]10.7086,[32]10.2419,[33]9.8871,[34]9.6029,[35]9.3962,[36]9.5205,[37]9.7559,[38]9.8816,[39]10.0292,
120
- save_imatrix: stored collected data after 40 chunks in gemma-2-27b-it-IMat-GGUF/imatrix.dat
121
- [40]10.1462,[41]10.2011,[42]10.6384,[43]10.9116,[44]11.2397,[45]11.4389,[46]11.2550,[47]11.0805,[48]11.2637,[49]11.4301,
122
- save_imatrix: stored collected data after 50 chunks in gemma-2-27b-it-IMat-GGUF/imatrix.dat
123
- [50]11.2782,[51]11.1894,[52]11.2327,[53]11.4036,[54]11.5967,[55]11.8117,[56]11.9157,[57]11.9185,[58]11.9380,[59]11.7810,
124
- save_imatrix: stored collected data after 60 chunks in gemma-2-27b-it-IMat-GGUF/imatrix.dat
125
- [60]11.6650,[61]11.5287,[62]11.4071,[63]11.4760,[64]11.5875,[65]11.4612,[66]11.4694,[67]11.4358,[68]11.4182,[69]11.3768,
126
- save_imatrix: stored collected data after 70 chunks in gemma-2-27b-it-IMat-GGUF/imatrix.dat
127
- [70]11.3167,[71]11.3025,[72]11.2862,[73]11.3380,[74]11.2888,[75]11.1897,[76]11.1629,[77]11.1599,[78]11.1381,[79]11.0667,
128
- save_imatrix: stored collected data after 80 chunks in gemma-2-27b-it-IMat-GGUF/imatrix.dat
129
- [80]11.1144,[81]11.1690,[82]11.1841,[83]11.2762,[84]11.2937,[85]11.1199,[86]11.0622,[87]10.9503,[88]10.9787,[89]10.9774,
130
- save_imatrix: stored collected data after 90 chunks in gemma-2-27b-it-IMat-GGUF/imatrix.dat
131
- [90]11.0439,[91]11.0167,[92]10.9757,[93]10.9302,[94]10.8609,[95]10.8306,[96]10.7705,[97]10.7326,[98]10.6794,[99]10.7099,
132
- save_imatrix: stored collected data after 100 chunks in gemma-2-27b-it-IMat-GGUF/imatrix.dat
133
- [100]10.7050,[101]10.8097,[102]10.8830,[103]10.9458,[104]11.0763,[105]11.1774,[106]11.1839,[107]11.1889,[108]11.1470,[109]11.1662,
134
- save_imatrix: stored collected data after 110 chunks in gemma-2-27b-it-IMat-GGUF/imatrix.dat
135
- [110]11.0690,[111]10.9087,[112]10.7355,[113]10.7965,[114]10.8346,[115]10.8238,[116]10.8008,[117]10.8403,[118]10.8670,[119]10.8845,
136
- save_imatrix: stored collected data after 120 chunks in gemma-2-27b-it-IMat-GGUF/imatrix.dat
137
- [120]10.8804,[121]10.8749,[122]10.8379,[123]10.8537,[124]10.9275,[125]11.0106,[126]11.0989,[127]11.1369,[128]11.1788,
138
- save_imatrix: stored collected data after 128 chunks in gemma-2-27b-it-IMat-GGUF/imatrix.dat
139
-
140
- llama_print_timings: load time = 4035.26 ms
141
- llama_print_timings: sample time = 0.00 ms / 1 runs ( 0.00 ms per token, inf tokens per second)
142
- llama_print_timings: prompt eval time = 226486.47 ms / 65536 tokens ( 3.46 ms per token, 289.36 tokens per second)
143
- llama_print_timings: eval time = 0.00 ms / 1 runs ( 0.00 ms per token, inf tokens per second)
144
- llama_print_timings: total time = 230157.77 ms / 65537 tokens
145
 
146
- Final estimate: PPL = 11.1788 +/- 0.20186
 
 
 
 
1
+ build: 3790 (5cb12f68) with cc (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 for x86_64-linux-gnu
2
+ llama_model_loader: loaded meta data with 39 key-value pairs and 508 tensors from gemma-2-27b-it-IMat-GGUF/gemma-2-27b-it.Q8_0.gguf.hardlink.gguf (version GGUF V3 (latest))
3
  llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output.
4
  llama_model_loader: - kv 0: general.architecture str = gemma2
5
+ llama_model_loader: - kv 1: general.type str = model
6
+ llama_model_loader: - kv 2: general.name str = Gemma 2 27b It
7
+ llama_model_loader: - kv 3: general.finetune str = it
8
+ llama_model_loader: - kv 4: general.basename str = gemma-2
9
+ llama_model_loader: - kv 5: general.size_label str = 27B
10
+ llama_model_loader: - kv 6: general.license str = gemma
11
+ llama_model_loader: - kv 7: general.base_model.count u32 = 1
12
+ llama_model_loader: - kv 8: general.base_model.0.name str = Gemma 2 27b
13
+ llama_model_loader: - kv 9: general.base_model.0.organization str = Google
14
+ llama_model_loader: - kv 10: general.base_model.0.repo_url str = https://huggingface.co/google/gemma-2...
15
+ llama_model_loader: - kv 11: general.tags arr[str,1] = ["text-generation"]
16
+ llama_model_loader: - kv 12: gemma2.context_length u32 = 8192
17
+ llama_model_loader: - kv 13: gemma2.embedding_length u32 = 4608
18
+ llama_model_loader: - kv 14: gemma2.block_count u32 = 46
19
+ llama_model_loader: - kv 15: gemma2.feed_forward_length u32 = 36864
20
+ llama_model_loader: - kv 16: gemma2.attention.head_count u32 = 32
21
+ llama_model_loader: - kv 17: gemma2.attention.head_count_kv u32 = 16
22
+ llama_model_loader: - kv 18: gemma2.attention.layer_norm_rms_epsilon f32 = 0.000001
23
+ llama_model_loader: - kv 19: gemma2.attention.key_length u32 = 128
24
+ llama_model_loader: - kv 20: gemma2.attention.value_length u32 = 128
25
+ llama_model_loader: - kv 21: general.file_type u32 = 7
26
+ llama_model_loader: - kv 22: gemma2.attn_logit_softcapping f32 = 50.000000
27
+ llama_model_loader: - kv 23: gemma2.final_logit_softcapping f32 = 30.000000
28
+ llama_model_loader: - kv 24: gemma2.attention.sliding_window u32 = 4096
29
+ llama_model_loader: - kv 25: tokenizer.ggml.model str = llama
30
+ llama_model_loader: - kv 26: tokenizer.ggml.pre str = default
31
+ llama_model_loader: - kv 27: tokenizer.ggml.tokens arr[str,256000] = ["<pad>", "<eos>", "<bos>", "<unk>", ...
32
+ llama_model_loader: - kv 28: tokenizer.ggml.scores arr[f32,256000] = [-1000.000000, -1000.000000, -1000.00...
33
+ llama_model_loader: - kv 29: tokenizer.ggml.token_type arr[i32,256000] = [3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, ...
34
+ llama_model_loader: - kv 30: tokenizer.ggml.bos_token_id u32 = 2
35
+ llama_model_loader: - kv 31: tokenizer.ggml.eos_token_id u32 = 1
36
+ llama_model_loader: - kv 32: tokenizer.ggml.unknown_token_id u32 = 3
37
+ llama_model_loader: - kv 33: tokenizer.ggml.padding_token_id u32 = 0
38
+ llama_model_loader: - kv 34: tokenizer.ggml.add_bos_token bool = true
39
+ llama_model_loader: - kv 35: tokenizer.ggml.add_eos_token bool = false
40
+ llama_model_loader: - kv 36: tokenizer.chat_template str = {{ bos_token }}{% if messages[0]['rol...
41
+ llama_model_loader: - kv 37: tokenizer.ggml.add_space_prefix bool = false
42
+ llama_model_loader: - kv 38: general.quantization_version u32 = 2
43
  llama_model_loader: - type f32: 185 tensors
44
  llama_model_loader: - type q8_0: 323 tensors
45
+ llm_load_vocab: special tokens cache size = 249
46
  llm_load_vocab: token to piece cache size = 1.6014 MB
47
  llm_load_print_meta: format = GGUF V3 (latest)
48
  llm_load_print_meta: arch = gemma2
49
  llm_load_print_meta: vocab type = SPM
50
  llm_load_print_meta: n_vocab = 256000
51
  llm_load_print_meta: n_merges = 0
52
+ llm_load_print_meta: vocab_only = 0
53
  llm_load_print_meta: n_ctx_train = 8192
54
  llm_load_print_meta: n_embd = 4608
55
+ llm_load_print_meta: n_layer = 46
56
  llm_load_print_meta: n_head = 32
57
  llm_load_print_meta: n_head_kv = 16
58
+ llm_load_print_meta: n_rot = 128
59
+ llm_load_print_meta: n_swa = 4096
60
  llm_load_print_meta: n_embd_head_k = 128
61
  llm_load_print_meta: n_embd_head_v = 128
62
  llm_load_print_meta: n_gqa = 2
 
82
  llm_load_print_meta: ssm_d_inner = 0
83
  llm_load_print_meta: ssm_d_state = 0
84
  llm_load_print_meta: ssm_dt_rank = 0
85
+ llm_load_print_meta: ssm_dt_b_c_rms = 0
86
  llm_load_print_meta: model type = 27B
87
  llm_load_print_meta: model ftype = Q8_0
88
  llm_load_print_meta: model params = 27.23 B
89
  llm_load_print_meta: model size = 26.94 GiB (8.50 BPW)
90
+ llm_load_print_meta: general.name = Gemma 2 27b It
91
  llm_load_print_meta: BOS token = 2 '<bos>'
92
  llm_load_print_meta: EOS token = 1 '<eos>'
93
  llm_load_print_meta: UNK token = 3 '<unk>'
94
  llm_load_print_meta: PAD token = 0 '<pad>'
95
  llm_load_print_meta: LF token = 227 '<0x0A>'
96
  llm_load_print_meta: EOT token = 107 '<end_of_turn>'
97
+ llm_load_print_meta: max token length = 48
98
  ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no
99
  ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
100
  ggml_cuda_init: found 1 CUDA devices:
 
116
  llama_new_context_with_model: KV self size = 184.00 MiB, K (f16): 92.00 MiB, V (f16): 92.00 MiB
117
  llama_new_context_with_model: CUDA_Host output buffer size = 0.98 MiB
118
  llama_new_context_with_model: CUDA0 compute buffer size = 1704.31 MiB
119
+ llama_new_context_with_model: CUDA_Host compute buffer size = 11.01 MiB
120
  llama_new_context_with_model: graph nodes = 1850
121
  llama_new_context_with_model: graph splits = 121
122
 
123
+ system_info: n_threads = 25 (n_threads_batch = 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 | RISCV_VECT = 0 | WASM_SIMD = 0 | BLAS = 1 | SSE3 = 1 | SSSE3 = 1 | VSX = 0 | MATMUL_INT8 = 0 | LLAMAFILE = 1 |
124
  compute_imatrix: tokenizing the input ..
125
+ compute_imatrix: tokenization took 125.601 ms
126
  compute_imatrix: computing over 128 chunks with batch_size 512
127
+ compute_imatrix: 2.33 seconds per pass - ETA 4.95 minutes
128
+ [1]6.5578,[2]4.3832,[3]3.8146,[4]4.7295,[5]4.8558,[6]4.2129,[7]4.5158,[8]4.7571,[9]4.9990,[10]4.4817,[11]4.6395,[12]5.0579,[13]5.5302,[14]5.7411,[15]6.1889,[16]6.4760,[17]6.6184,[18]6.9135,[19]6.6034,[20]6.7227,[21]6.8527,[22]6.8112,[23]6.9548,[24]6.9825,[25]7.1554,[26]6.9450,[27]7.1144,[28]7.2338,[29]7.1511,[30]7.1380,[31]6.7390,[32]6.5218,[33]6.4390,[34]6.3315,[35]6.2964,[36]6.3155,[37]6.3346,[38]6.3824,[39]6.5202,[40]6.6510,[41]6.7730,[42]7.0004,[43]7.2346,[44]7.4522,[45]7.5993,[46]7.4966,[47]7.5276,[48]7.6901,[49]7.8164,[50]7.6607,[51]7.6948,[52]7.7278,[53]7.8302,[54]7.9708,[55]8.0723,[56]8.1382,[57]8.1527,[58]8.1844,[59]8.0526,[60]7.9599,[61]7.8446,[62]7.8175,[63]7.8454,[64]7.8452,[65]7.8189,[66]7.8290,[67]7.7794,[68]7.7266,[69]7.7449,[70]7.7104,[71]7.6979,[72]7.7069,[73]7.6929,[74]7.6497,[75]7.6120,[76]7.6072,[77]7.6029,[78]7.5971,[79]7.5538,[80]7.6036,[81]7.6355,[82]7.6205,[83]7.6298,[84]7.6690,[85]7.5607,[86]7.5246,[87]7.4642,[88]7.4749,[89]7.5126,[90]7.5311,[91]7.4788,[92]7.4214,[93]7.3542,[94]7.2928,[95]7.2487,[96]7.1886,[97]7.1361,[98]7.0910,[99]7.1216,[100]7.1453,[101]7.2155,[102]7.2828,[103]7.3533,[104]7.4761,[105]7.5655,[106]7.5829,[107]7.6048,[108]7.6214,[109]7.5943,[110]7.5657,[111]7.4757,[112]7.3784,[113]7.4231,[114]7.4466,[115]7.4463,[116]7.4460,[117]7.4808,[118]7.4960,[119]7.5066,[120]7.5191,[121]7.5362,[122]7.5135,[123]7.5547,[124]7.6026,[125]7.6386,[126]7.7025,[127]7.7520,[128]7.7917,
129
+ Final estimate: PPL = 7.7917 +/- 0.11823
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
130
 
131
+ llama_perf_context_print: load time = 5557.94 ms
132
+ llama_perf_context_print: prompt eval time = 238949.63 ms / 65536 tokens ( 3.65 ms per token, 274.27 tokens per second)
133
+ llama_perf_context_print: eval time = 0.00 ms / 1 runs ( 0.00 ms per token, inf tokens per second)
134
+ llama_perf_context_print: total time = 243783.26 ms / 65537 tokens