Hugging Face's logo

legraphista
/
DeepSeek-Coder-V2-Instruct-0724-IMat-GGUF

Text Generation
GGUF
quantized
GGUF
quantization
imat
imatrix
static
8bit
6bit
5bit
4bit
3bit
2bit
1bit
conversational
Model card Files Community
legraphista commited on
Commit
1e414c4
•
1 Parent(s): 653f560

Upload imatrix.log with huggingface_hub

Browse files
Files changed (1) hide show
  1. imatrix.log +178 -0
imatrix.log ADDED
@@ -0,0 +1,178 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ llama_model_loader: loaded meta data with 51 key-value pairs and 959 tensors from DeepSeek-Coder-V2-Instruct-0724-IMat-GGUF/DeepSeek-Coder-V2-Instruct-0724.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 = deepseek2
4
+ llama_model_loader: - kv 1: general.type str = model
5
+ llama_model_loader: - kv 2: general.name str = DeepSeek Coder V2 Instruct 0724
6
+ llama_model_loader: - kv 3: general.version str = 0724
7
+ llama_model_loader: - kv 4: general.finetune str = Instruct
8
+ llama_model_loader: - kv 5: general.basename str = DeepSeek-Coder-V2
9
+ llama_model_loader: - kv 6: general.size_label str = 160x14B
10
+ llama_model_loader: - kv 7: general.license str = other
11
+ llama_model_loader: - kv 8: general.license.name str = deepseek-license
12
+ llama_model_loader: - kv 9: general.license.link str = LICENSE
13
+ llama_model_loader: - kv 10: general.base_model.count u32 = 1
14
+ llama_model_loader: - kv 11: general.base_model.0.name str = DeepSeek Coder V2 Base
15
+ llama_model_loader: - kv 12: general.base_model.0.organization str = Deepseek Ai
16
+ llama_model_loader: - kv 13: general.base_model.0.repo_url str = https://huggingface.co/deepseek-ai/De...
17
+ llama_model_loader: - kv 14: deepseek2.block_count u32 = 60
18
+ llama_model_loader: - kv 15: deepseek2.context_length u32 = 163840
19
+ llama_model_loader: - kv 16: deepseek2.embedding_length u32 = 5120
20
+ llama_model_loader: - kv 17: deepseek2.feed_forward_length u32 = 12288
21
+ llama_model_loader: - kv 18: deepseek2.attention.head_count u32 = 128
22
+ llama_model_loader: - kv 19: deepseek2.attention.head_count_kv u32 = 128
23
+ llama_model_loader: - kv 20: deepseek2.rope.freq_base f32 = 10000.000000
24
+ llama_model_loader: - kv 21: deepseek2.attention.layer_norm_rms_epsilon f32 = 0.000001
25
+ llama_model_loader: - kv 22: deepseek2.expert_used_count u32 = 6
26
+ llama_model_loader: - kv 23: general.file_type u32 = 7
27
+ llama_model_loader: - kv 24: deepseek2.leading_dense_block_count u32 = 1
28
+ llama_model_loader: - kv 25: deepseek2.vocab_size u32 = 102400
29
+ llama_model_loader: - kv 26: deepseek2.attention.q_lora_rank u32 = 1536
30
+ llama_model_loader: - kv 27: deepseek2.attention.kv_lora_rank u32 = 512
31
+ llama_model_loader: - kv 28: deepseek2.attention.key_length u32 = 192
32
+ llama_model_loader: - kv 29: deepseek2.attention.value_length u32 = 128
33
+ llama_model_loader: - kv 30: deepseek2.expert_feed_forward_length u32 = 1536
34
+ llama_model_loader: - kv 31: deepseek2.expert_count u32 = 160
35
+ llama_model_loader: - kv 32: deepseek2.expert_shared_count u32 = 2
36
+ llama_model_loader: - kv 33: deepseek2.expert_weights_scale f32 = 16.000000
37
+ llama_model_loader: - kv 34: deepseek2.rope.dimension_count u32 = 64
38
+ llama_model_loader: - kv 35: deepseek2.rope.scaling.type str = yarn
39
+ llama_model_loader: - kv 36: deepseek2.rope.scaling.factor f32 = 40.000000
40
+ llama_model_loader: - kv 37: deepseek2.rope.scaling.original_context_length u32 = 4096
41
+ llama_model_loader: - kv 38: deepseek2.rope.scaling.yarn_log_multiplier f32 = 0.100000
42
+ llama_model_loader: - kv 39: tokenizer.ggml.model str = gpt2
43
+ llama_model_loader: - kv 40: tokenizer.ggml.pre str = deepseek-llm
44
+ llama_model_loader: - kv 41: tokenizer.ggml.tokens arr[str,102400] = ["!", "\"", "#", "$", "%", "&", "'", ...
45
+ llama_model_loader: - kv 42: tokenizer.ggml.token_type arr[i32,102400] = [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, ...
46
+ llama_model_loader: - kv 43: tokenizer.ggml.merges arr[str,99757] = ["Ġ Ġ", "Ġ t", "Ġ a", "i n", "h e...
47
+ llama_model_loader: - kv 44: tokenizer.ggml.bos_token_id u32 = 100000
48
+ llama_model_loader: - kv 45: tokenizer.ggml.eos_token_id u32 = 100001
49
+ llama_model_loader: - kv 46: tokenizer.ggml.padding_token_id u32 = 100001
50
+ llama_model_loader: - kv 47: tokenizer.ggml.add_bos_token bool = true
51
+ llama_model_loader: - kv 48: tokenizer.ggml.add_eos_token bool = false
52
+ llama_model_loader: - kv 49: tokenizer.chat_template str = {% if not add_generation_prompt is de...
53
+ llama_model_loader: - kv 50: general.quantization_version u32 = 2
54
+ llama_model_loader: - type f32: 300 tensors
55
+ llama_model_loader: - type q8_0: 659 tensors
56
+ llm_load_vocab: special tokens cache size = 18
57
+ llm_load_vocab: token to piece cache size = 0.6411 MB
58
+ llm_load_print_meta: format = GGUF V3 (latest)
59
+ llm_load_print_meta: arch = deepseek2
60
+ llm_load_print_meta: vocab type = BPE
61
+ llm_load_print_meta: n_vocab = 102400
62
+ llm_load_print_meta: n_merges = 99757
63
+ llm_load_print_meta: vocab_only = 0
64
+ llm_load_print_meta: n_ctx_train = 163840
65
+ llm_load_print_meta: n_embd = 5120
66
+ llm_load_print_meta: n_layer = 60
67
+ llm_load_print_meta: n_head = 128
68
+ llm_load_print_meta: n_head_kv = 128
69
+ llm_load_print_meta: n_rot = 64
70
+ llm_load_print_meta: n_swa = 0
71
+ llm_load_print_meta: n_embd_head_k = 192
72
+ llm_load_print_meta: n_embd_head_v = 128
73
+ llm_load_print_meta: n_gqa = 1
74
+ llm_load_print_meta: n_embd_k_gqa = 24576
75
+ llm_load_print_meta: n_embd_v_gqa = 16384
76
+ llm_load_print_meta: f_norm_eps = 0.0e+00
77
+ llm_load_print_meta: f_norm_rms_eps = 1.0e-06
78
+ llm_load_print_meta: f_clamp_kqv = 0.0e+00
79
+ llm_load_print_meta: f_max_alibi_bias = 0.0e+00
80
+ llm_load_print_meta: f_logit_scale = 0.0e+00
81
+ llm_load_print_meta: n_ff = 12288
82
+ llm_load_print_meta: n_expert = 160
83
+ llm_load_print_meta: n_expert_used = 6
84
+ llm_load_print_meta: causal attn = 1
85
+ llm_load_print_meta: pooling type = 0
86
+ llm_load_print_meta: rope type = 0
87
+ llm_load_print_meta: rope scaling = yarn
88
+ llm_load_print_meta: freq_base_train = 10000.0
89
+ llm_load_print_meta: freq_scale_train = 0.025
90
+ llm_load_print_meta: n_ctx_orig_yarn = 4096
91
+ llm_load_print_meta: rope_finetuned = unknown
92
+ llm_load_print_meta: ssm_d_conv = 0
93
+ llm_load_print_meta: ssm_d_inner = 0
94
+ llm_load_print_meta: ssm_d_state = 0
95
+ llm_load_print_meta: ssm_dt_rank = 0
96
+ llm_load_print_meta: ssm_dt_b_c_rms = 0
97
+ llm_load_print_meta: model type = 236B
98
+ llm_load_print_meta: model ftype = Q8_0
99
+ llm_load_print_meta: model params = 235.74 B
100
+ llm_load_print_meta: model size = 233.41 GiB (8.50 BPW)
101
+ llm_load_print_meta: general.name = DeepSeek Coder V2 Instruct 0724
102
+ llm_load_print_meta: BOS token = 100000 '<｜begin▁of▁sentence｜>'
103
+ llm_load_print_meta: EOS token = 100001 '<｜end▁of▁sentence｜>'
104
+ llm_load_print_meta: PAD token = 100001 '<｜end▁of▁sentence｜>'
105
+ llm_load_print_meta: LF token = 126 'Ä'
106
+ llm_load_print_meta: max token length = 256
107
+ llm_load_print_meta: n_layer_dense_lead = 1
108
+ llm_load_print_meta: n_lora_q = 1536
109
+ llm_load_print_meta: n_lora_kv = 512
110
+ llm_load_print_meta: n_ff_exp = 1536
111
+ llm_load_print_meta: n_expert_shared = 2
112
+ llm_load_print_meta: expert_weights_scale = 16.0
113
+ llm_load_print_meta: rope_yarn_log_mul = 0.1000
114
+ ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no
115
+ ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
116
+ ggml_cuda_init: found 1 CUDA devices:
117
+ Device 0: NVIDIA GeForce RTX 4090, compute capability 8.9, VMM: yes
118
+ llm_load_tensors: ggml ctx size = 0.80 MiB
119
+ llm_load_tensors: offloading 5 repeating layers to GPU
120
+ llm_load_tensors: offloaded 5/61 layers to GPU
121
+ llm_load_tensors: CPU buffer size = 218873.36 MiB
122
+ llm_load_tensors: CUDA0 buffer size = 20135.96 MiB
123
+ ....................................................................................................
124
+ llama_new_context_with_model: n_ctx = 512
125
+ llama_new_context_with_model: n_batch = 512
126
+ llama_new_context_with_model: n_ubatch = 512
127
+ llama_new_context_with_model: flash_attn = 0
128
+ llama_new_context_with_model: freq_base = 10000.0
129
+ llama_new_context_with_model: freq_scale = 0.025
130
+ llama_kv_cache_init: CUDA_Host KV buffer size = 2200.00 MiB
131
+ llama_kv_cache_init: CUDA0 KV buffer size = 200.00 MiB
132
+ llama_new_context_with_model: KV self size = 2400.00 MiB, K (f16): 1440.00 MiB, V (f16): 960.00 MiB
133
+ llama_new_context_with_model: CUDA_Host output buffer size = 0.39 MiB
134
+ llama_new_context_with_model: CUDA0 compute buffer size = 1422.00 MiB
135
+ llama_new_context_with_model: CUDA_Host compute buffer size = 81.01 MiB
136
+ llama_new_context_with_model: graph nodes = 4480
137
+ llama_new_context_with_model: graph splits = 990
138
+
139
+ 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 | WASM_SIMD = 0 | BLAS = 1 | SSE3 = 1 | SSSE3 = 1 | VSX = 0 | MATMUL_INT8 = 0 | LLAMAFILE = 1 |
140
+ compute_imatrix: tokenizing the input ..
141
+ compute_imatrix: tokenization took 526.799 ms
142
+ compute_imatrix: computing over 139 chunks with batch_size 512
143
+ compute_imatrix: 90.65 seconds per pass - ETA 3 hours 30.00 minutes
144
+ [1]5.7772,[2]3.8835,[3]3.8059,[4]4.1672,[5]4.1086,[6]3.9026,[7]4.1099,[8]3.9733,[9]4.4323,
145
+ save_imatrix: stored collected data after 10 chunks in DeepSeek-Coder-V2-Instruct-0724-IMat-GGUF/imatrix.dat
146
+ [10]4.6201,[11]4.3030,[12]4.5590,[13]4.8871,[14]5.1335,[15]5.2601,[16]5.4875,[17]5.6677,[18]5.7927,[19]5.8962,
147
+ save_imatrix: stored collected data after 20 chunks in DeepSeek-Coder-V2-Instruct-0724-IMat-GGUF/imatrix.dat
148
+ [20]5.6232,[21]5.6381,[22]5.5802,[23]5.6274,[24]5.5363,[25]5.6976,[26]5.6137,[27]5.7519,[28]5.6031,[29]5.3284,
149
+ save_imatrix: stored collected data after 30 chunks in DeepSeek-Coder-V2-Instruct-0724-IMat-GGUF/imatrix.dat
150
+ [30]5.1049,[31]5.0415,[32]4.9882,[33]4.8558,[34]4.6419,[35]4.4657,[36]4.4172,[37]4.3647,[38]4.3780,[39]4.3155,
151
+ save_imatrix: stored collected data after 40 chunks in DeepSeek-Coder-V2-Instruct-0724-IMat-GGUF/imatrix.dat
152
+ [40]4.2715,[41]4.2049,[42]4.1509,[43]4.1936,[44]4.2619,[45]4.3487,[46]4.3656,[47]4.2404,[48]4.1211,[49]4.0191,
153
+ save_imatrix: stored collected data after 50 chunks in DeepSeek-Coder-V2-Instruct-0724-IMat-GGUF/imatrix.dat
154
+ [50]3.9192,[51]3.9238,[52]3.8966,[53]3.9838,[54]4.0627,[55]4.1495,[56]4.1101,[57]4.1283,[58]4.1489,[59]4.2108,
155
+ save_imatrix: stored collected data after 60 chunks in DeepSeek-Coder-V2-Instruct-0724-IMat-GGUF/imatrix.dat
156
+ [60]4.2794,[61]4.3451,[62]4.3897,[63]4.4198,[64]4.4604,[65]4.4913,[66]4.4892,[67]4.4981,[68]4.5029,[69]4.5437,
157
+ save_imatrix: stored collected data after 70 chunks in DeepSeek-Coder-V2-Instruct-0724-IMat-GGUF/imatrix.dat
158
+ [70]4.5644,[71]4.5709,[72]4.6022,[73]4.6110,[74]4.6151,[75]4.6210,[76]4.6322,[77]4.6445,[78]4.6694,[79]4.6579,
159
+ save_imatrix: stored collected data after 80 chunks in DeepSeek-Coder-V2-Instruct-0724-IMat-GGUF/imatrix.dat
160
+ [80]4.6792,[81]4.6795,[82]4.6854,[83]4.6823,[84]4.6838,[85]4.6807,[86]4.6792,[87]4.6723,[88]4.6972,[89]4.7218,
161
+ save_imatrix: stored collected data after 90 chunks in DeepSeek-Coder-V2-Instruct-0724-IMat-GGUF/imatrix.dat
162
+ [90]4.7272,[91]4.7600,[92]4.7942,[93]4.7672,[94]4.7706,[95]4.7531,[96]4.7771,[97]4.7976,[98]4.7951,[99]4.7560,
163
+ save_imatrix: stored collected data after 100 chunks in DeepSeek-Coder-V2-Instruct-0724-IMat-GGUF/imatrix.dat
164
+ [100]4.7169,[101]4.6779,[102]4.6376,[103]4.5994,[104]4.5669,[105]4.5309,[106]4.4965,[107]4.4640,[108]4.4398,[109]4.4545,
165
+ save_imatrix: stored collected data after 110 chunks in DeepSeek-Coder-V2-Instruct-0724-IMat-GGUF/imatrix.dat
166
+ [110]4.4866,[111]4.5306,[112]4.5749,[113]4.6178,[114]4.6884,[115]4.7267,[116]4.7498,[117]4.7678,[118]4.7937,[119]4.7927,
167
+ save_imatrix: stored collected data after 120 chunks in DeepSeek-Coder-V2-Instruct-0724-IMat-GGUF/imatrix.dat
168
+ [120]4.7703,[121]4.7169,[122]4.6675,[123]4.6978,[124]4.7296,[125]4.7339,[126]4.7399,[127]4.7519,[128]4.7785,[129]4.7832,
169
+ save_imatrix: stored collected data after 130 chunks in DeepSeek-Coder-V2-Instruct-0724-IMat-GGUF/imatrix.dat
170
+ [130]4.7964,[131]4.8182,[132]4.8148,[133]4.8068,[134]4.7571,[135]4.7070,[136]4.7102,[137]4.6635,[138]4.6222,[139]4.5770,
171
+ save_imatrix: stored collected data after 139 chunks in DeepSeek-Coder-V2-Instruct-0724-IMat-GGUF/imatrix.dat
172
+
173
+ llama_perf_print: load time = 194329.99 ms
174
+ llama_perf_print: prompt eval time = 15123291.91 ms / 71168 tokens ( 212.50 ms per token, 4.71 tokens per second)
175
+ llama_perf_print: eval time = 0.00 ms / 1 runs ( 0.00 ms per token, inf tokens per second)
176
+ llama_perf_print: total time = 15252664.04 ms / 71169 tokens
177
+
178
+ Final estimate: PPL = 4.5770 +/- 0.05779