Hugging Face's logo

legraphista
/
Yi-Coder-9B-IMat-GGUF

Text Generation
GGUF
quantized
GGUF
quantization
imat
imatrix
static
16bit
8bit
6bit
5bit
4bit
3bit
2bit
1bit
Model card Files Community
legraphista commited on
Commit
4e05306
•
1 Parent(s): 6e0dcd0

Upload imatrix.log with huggingface_hub

Browse files
Files changed (1) hide show
  1. imatrix.log +154 -0
imatrix.log ADDED
@@ -0,0 +1,154 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ llama_model_loader: loaded meta data with 29 key-value pairs and 435 tensors from Yi-Coder-9B-IMat-GGUF/Yi-Coder-9B.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 = llama
4
+ llama_model_loader: - kv 1: general.type str = model
5
+ llama_model_loader: - kv 2: general.name str = Yi Coder 9B
6
+ llama_model_loader: - kv 3: general.basename str = Yi-Coder
7
+ llama_model_loader: - kv 4: general.size_label str = 9B
8
+ llama_model_loader: - kv 5: general.license str = apache-2.0
9
+ llama_model_loader: - kv 6: llama.block_count u32 = 48
10
+ llama_model_loader: - kv 7: llama.context_length u32 = 131072
11
+ llama_model_loader: - kv 8: llama.embedding_length u32 = 4096
12
+ llama_model_loader: - kv 9: llama.feed_forward_length u32 = 11008
13
+ llama_model_loader: - kv 10: llama.attention.head_count u32 = 32
14
+ llama_model_loader: - kv 11: llama.attention.head_count_kv u32 = 4
15
+ llama_model_loader: - kv 12: llama.rope.freq_base f32 = 10000000.000000
16
+ llama_model_loader: - kv 13: llama.attention.layer_norm_rms_epsilon f32 = 0.000010
17
+ llama_model_loader: - kv 14: general.file_type u32 = 7
18
+ llama_model_loader: - kv 15: llama.vocab_size u32 = 64000
19
+ llama_model_loader: - kv 16: llama.rope.dimension_count u32 = 128
20
+ llama_model_loader: - kv 17: tokenizer.ggml.model str = llama
21
+ llama_model_loader: - kv 18: tokenizer.ggml.pre str = default
22
+ llama_model_loader: - kv 19: tokenizer.ggml.tokens arr[str,64000] = ["<unk>", "<|startoftext|>", "<|endof...
23
+ llama_model_loader: - kv 20: tokenizer.ggml.scores arr[f32,64000] = [-1000.000000, -1000.000000, -1000.00...
24
+ llama_model_loader: - kv 21: tokenizer.ggml.token_type arr[i32,64000] = [3, 3, 3, 3, 3, 3, 1, 1, 1, 3, 3, 3, ...
25
+ llama_model_loader: - kv 22: tokenizer.ggml.bos_token_id u32 = 1
26
+ llama_model_loader: - kv 23: tokenizer.ggml.eos_token_id u32 = 2
27
+ llama_model_loader: - kv 24: tokenizer.ggml.unknown_token_id u32 = 0
28
+ llama_model_loader: - kv 25: tokenizer.ggml.padding_token_id u32 = 0
29
+ llama_model_loader: - kv 26: tokenizer.ggml.add_bos_token bool = false
30
+ llama_model_loader: - kv 27: tokenizer.ggml.add_eos_token bool = false
31
+ llama_model_loader: - kv 28: general.quantization_version u32 = 2
32
+ llama_model_loader: - type f32: 97 tensors
33
+ llama_model_loader: - type q8_0: 338 tensors
34
+ llm_load_vocab: special tokens cache size = 11
35
+ llm_load_vocab: token to piece cache size = 0.3834 MB
36
+ llm_load_print_meta: format = GGUF V3 (latest)
37
+ llm_load_print_meta: arch = llama
38
+ llm_load_print_meta: vocab type = SPM
39
+ llm_load_print_meta: n_vocab = 64000
40
+ llm_load_print_meta: n_merges = 0
41
+ llm_load_print_meta: vocab_only = 0
42
+ llm_load_print_meta: n_ctx_train = 131072
43
+ llm_load_print_meta: n_embd = 4096
44
+ llm_load_print_meta: n_layer = 48
45
+ llm_load_print_meta: n_head = 32
46
+ llm_load_print_meta: n_head_kv = 4
47
+ llm_load_print_meta: n_rot = 128
48
+ llm_load_print_meta: n_swa = 0
49
+ llm_load_print_meta: n_embd_head_k = 128
50
+ llm_load_print_meta: n_embd_head_v = 128
51
+ llm_load_print_meta: n_gqa = 8
52
+ llm_load_print_meta: n_embd_k_gqa = 512
53
+ llm_load_print_meta: n_embd_v_gqa = 512
54
+ llm_load_print_meta: f_norm_eps = 0.0e+00
55
+ llm_load_print_meta: f_norm_rms_eps = 1.0e-05
56
+ llm_load_print_meta: f_clamp_kqv = 0.0e+00
57
+ llm_load_print_meta: f_max_alibi_bias = 0.0e+00
58
+ llm_load_print_meta: f_logit_scale = 0.0e+00
59
+ llm_load_print_meta: n_ff = 11008
60
+ llm_load_print_meta: n_expert = 0
61
+ llm_load_print_meta: n_expert_used = 0
62
+ llm_load_print_meta: causal attn = 1
63
+ llm_load_print_meta: pooling type = 0
64
+ llm_load_print_meta: rope type = 0
65
+ llm_load_print_meta: rope scaling = linear
66
+ llm_load_print_meta: freq_base_train = 10000000.0
67
+ llm_load_print_meta: freq_scale_train = 1
68
+ llm_load_print_meta: n_ctx_orig_yarn = 131072
69
+ llm_load_print_meta: rope_finetuned = unknown
70
+ llm_load_print_meta: ssm_d_conv = 0
71
+ llm_load_print_meta: ssm_d_inner = 0
72
+ llm_load_print_meta: ssm_d_state = 0
73
+ llm_load_print_meta: ssm_dt_rank = 0
74
+ llm_load_print_meta: ssm_dt_b_c_rms = 0
75
+ llm_load_print_meta: model type = 34B
76
+ llm_load_print_meta: model ftype = Q8_0
77
+ llm_load_print_meta: model params = 8.83 B
78
+ llm_load_print_meta: model size = 8.74 GiB (8.50 BPW)
79
+ llm_load_print_meta: general.name = Yi Coder 9B
80
+ llm_load_print_meta: BOS token = 1 '<|startoftext|>'
81
+ llm_load_print_meta: EOS token = 2 '<|endoftext|>'
82
+ llm_load_print_meta: UNK token = 0 '<unk>'
83
+ llm_load_print_meta: PAD token = 0 '<unk>'
84
+ llm_load_print_meta: LF token = 315 '<0x0A>'
85
+ llm_load_print_meta: EOT token = 2 '<|endoftext|>'
86
+ llm_load_print_meta: max token length = 48
87
+ ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no
88
+ ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
89
+ ggml_cuda_init: found 1 CUDA devices:
90
+ Device 0: NVIDIA GeForce RTX 4090, compute capability 8.9, VMM: yes
91
+ llm_load_tensors: ggml ctx size = 0.41 MiB
92
+ llm_load_tensors: offloading 48 repeating layers to GPU
93
+ llm_load_tensors: offloading non-repeating layers to GPU
94
+ llm_load_tensors: offloaded 49/49 layers to GPU
95
+ llm_load_tensors: CPU buffer size = 265.62 MiB
96
+ llm_load_tensors: CUDA0 buffer size = 8682.16 MiB
97
+ .................................................................................................
98
+ llama_new_context_with_model: n_ctx = 512
99
+ llama_new_context_with_model: n_batch = 512
100
+ llama_new_context_with_model: n_ubatch = 512
101
+ llama_new_context_with_model: flash_attn = 0
102
+ llama_new_context_with_model: freq_base = 10000000.0
103
+ llama_new_context_with_model: freq_scale = 1
104
+ llama_kv_cache_init: CUDA0 KV buffer size = 48.00 MiB
105
+ llama_new_context_with_model: KV self size = 48.00 MiB, K (f16): 24.00 MiB, V (f16): 24.00 MiB
106
+ llama_new_context_with_model: CUDA_Host output buffer size = 0.24 MiB
107
+ llama_new_context_with_model: CUDA0 compute buffer size = 133.00 MiB
108
+ llama_new_context_with_model: CUDA_Host compute buffer size = 9.01 MiB
109
+ llama_new_context_with_model: graph nodes = 1542
110
+ llama_new_context_with_model: graph splits = 2
111
+
112
+ 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 |
113
+ compute_imatrix: tokenizing the input ..
114
+ compute_imatrix: tokenization took 95.343 ms
115
+ compute_imatrix: computing over 146 chunks with batch_size 512
116
+ compute_imatrix: 0.79 seconds per pass - ETA 1.90 minutes
117
+ [1]7.2880,[2]5.1883,[3]5.2856,[4]5.5969,[5]5.5395,[6]5.8015,[7]4.9470,[8]5.5094,[9]5.4777,
118
+ save_imatrix: stored collected data after 10 chunks in Yi-Coder-9B-IMat-GGUF/imatrix.dat
119
+ [10]5.9434,[11]5.9764,[12]5.4793,[13]5.8090,[14]6.3599,[15]6.5982,[16]7.0200,[17]7.3588,[18]7.4586,[19]7.5772,
120
+ save_imatrix: stored collected data after 20 chunks in Yi-Coder-9B-IMat-GGUF/imatrix.dat
121
+ [20]7.8653,[21]7.5142,[22]7.4258,[23]7.5155,[24]7.5329,[25]7.5233,[26]7.2979,[27]7.5016,[28]7.6617,[29]7.9083,
122
+ save_imatrix: stored collected data after 30 chunks in Yi-Coder-9B-IMat-GGUF/imatrix.dat
123
+ [30]7.9808,[31]8.1886,[32]8.4090,[33]8.3950,[34]8.2353,[35]7.9854,[36]7.5750,[37]7.2076,[38]7.1694,[39]7.1115,
124
+ save_imatrix: stored collected data after 40 chunks in Yi-Coder-9B-IMat-GGUF/imatrix.dat
125
+ [40]7.0891,[41]6.8756,[42]6.6935,[43]6.5613,[44]6.3987,[45]6.2618,[46]6.2461,[47]6.3318,[48]6.4427,[49]6.5706,
126
+ save_imatrix: stored collected data after 50 chunks in Yi-Coder-9B-IMat-GGUF/imatrix.dat
127
+ [50]6.7226,[51]6.9875,[52]7.2196,[53]7.4019,[54]7.5350,[55]7.5759,[56]7.5082,[57]7.6188,[58]7.6928,[59]7.7780,
128
+ save_imatrix: stored collected data after 60 chunks in Yi-Coder-9B-IMat-GGUF/imatrix.dat
129
+ [60]7.7053,[61]7.6323,[62]7.6596,[63]7.7951,[64]7.9040,[65]8.0036,[66]8.0511,[67]8.1005,[68]8.1527,[69]8.1953,
130
+ save_imatrix: stored collected data after 70 chunks in Yi-Coder-9B-IMat-GGUF/imatrix.dat
131
+ [70]8.0910,[71]8.0134,[72]7.9389,[73]7.8956,[74]7.9284,[75]7.9880,[76]7.9779,[77]7.9928,[78]7.9899,[79]7.9561,
132
+ save_imatrix: stored collected data after 80 chunks in Yi-Coder-9B-IMat-GGUF/imatrix.dat
133
+ [80]7.9137,[81]7.8572,[82]7.8692,[83]7.8375,[84]7.8118,[85]7.8247,[86]7.7957,[87]7.7487,[88]7.7099,[89]7.7192,
134
+ save_imatrix: stored collected data after 90 chunks in Yi-Coder-9B-IMat-GGUF/imatrix.dat
135
+ [90]7.6908,[91]7.6689,[92]7.5977,[93]7.5844,[94]7.6399,[95]7.6387,[96]7.6007,[97]7.6168,[98]7.6299,[99]7.6494,
136
+ save_imatrix: stored collected data after 100 chunks in Yi-Coder-9B-IMat-GGUF/imatrix.dat
137
+ [100]7.5541,[101]7.6011,[102]7.6154,[103]7.6375,[104]7.6594,[105]7.6740,[106]7.6198,[107]7.5690,[108]7.5181,[109]7.4570,
138
+ save_imatrix: stored collected data after 110 chunks in Yi-Coder-9B-IMat-GGUF/imatrix.dat
139
+ [110]7.4035,[111]7.3519,[112]7.3042,[113]7.2572,[114]7.2227,[115]7.2385,[116]7.2824,[117]7.3530,[118]7.4214,[119]7.4886,
140
+ save_imatrix: stored collected data after 120 chunks in Yi-Coder-9B-IMat-GGUF/imatrix.dat
141
+ [120]7.6025,[121]7.6782,[122]7.7016,[123]7.7137,[124]7.6705,[125]7.6765,[126]7.6478,[127]7.5651,[128]7.4809,[129]7.4183,
142
+ save_imatrix: stored collected data after 130 chunks in Yi-Coder-9B-IMat-GGUF/imatrix.dat
143
+ [130]7.4634,[131]7.4627,[132]7.4897,[133]7.5171,[134]7.5629,[135]7.5906,[136]7.6044,[137]7.6149,[138]7.6157,[139]7.5994,
144
+ save_imatrix: stored collected data after 140 chunks in Yi-Coder-9B-IMat-GGUF/imatrix.dat
145
+ [140]7.6620,[141]7.7203,[142]7.7764,[143]7.8331,[144]7.8902,[145]7.9510,[146]7.9918,
146
+ save_imatrix: stored collected data after 146 chunks in Yi-Coder-9B-IMat-GGUF/imatrix.dat
147
+
148
+ llama_print_timings: load time = 2336.06 ms
149
+ llama_print_timings: sample time = 0.00 ms / 1 runs ( 0.00 ms per token, inf tokens per second)
150
+ llama_print_timings: prompt eval time = 108543.70 ms / 74752 tokens ( 1.45 ms per token, 688.68 tokens per second)
151
+ llama_print_timings: eval time = 0.00 ms / 1 runs ( 0.00 ms per token, inf tokens per second)
152
+ llama_print_timings: total time = 110682.85 ms / 74753 tokens
153
+
154
+ Final estimate: PPL = 7.9918 +/- 0.11014