Hugging Face's logo

amd
/
Llama-2-70b-chat-hf-WMXFP4-AMXFP4-KVFP8-Scale-UINT8-MLPerf-GPTQ

Safetensors
llama
8-bit precision
quark
Model card Files
xet
Community
1
linzhao-amd commited on
Commit
1d3d290
·
verified ·
1 Parent(s): 912ce00

upload gptq

Browse files
Files changed (14) hide show
  1. config.json +303 -0
  2. generation_config.json +10 -0
  3. model-00001-of-00008.safetensors +3 -0
  4. model-00002-of-00008.safetensors +3 -0
  5. model-00003-of-00008.safetensors +3 -0
  6. model-00004-of-00008.safetensors +3 -0
  7. model-00005-of-00008.safetensors +3 -0
  8. model-00006-of-00008.safetensors +3 -0
  9. model-00007-of-00008.safetensors +3 -0
  10. model-00008-of-00008.safetensors +3 -0
  11. model.safetensors.index.json +0 -0
  12. special_tokens_map.json +24 -0
  13. tokenizer.model +3 -0
  14. tokenizer_config.json +45 -0
config.json ADDED
@@ -0,0 +1,303 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "LlamaForCausalLM"
4
+ ],
5
+ "attention_bias": false,
6
+ "attention_dropout": 0.0,
7
+ "bos_token_id": 1,
8
+ "eos_token_id": 2,
9
+ "head_dim": 128,
10
+ "hidden_act": "silu",
11
+ "hidden_size": 8192,
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 28672,
14
+ "max_position_embeddings": 4096,
15
+ "mlp_bias": false,
16
+ "model_type": "llama",
17
+ "num_attention_heads": 64,
18
+ "num_hidden_layers": 80,
19
+ "num_key_value_heads": 8,
20
+ "pretraining_tp": 1,
21
+ "quantization_config": {
22
+ "algo_config": [
23
+ {
24
+ "block_size": 128,
25
+ "damp_percent": 0.01,
26
+ "desc_act": true,
27
+ "inside_layer_modules": [
28
+ "self_attn.k_proj",
29
+ "self_attn.v_proj",
30
+ "self_attn.q_proj",
31
+ "self_attn.o_proj",
32
+ "mlp.up_proj",
33
+ "mlp.gate_proj",
34
+ "mlp.down_proj"
35
+ ],
36
+ "model_decoder_layers": "model.layers",
37
+ "name": "gptq",
38
+ "static_groups": true
39
+ }
40
+ ],
41
+ "exclude": [
42
+ "lm_head"
43
+ ],
44
+ "export": {
45
+ "kv_cache_group": [
46
+ "*k_proj",
47
+ "*v_proj"
48
+ ],
49
+ "min_kv_scale": 0.0,
50
+ "pack_method": "reorder",
51
+ "weight_format": "real_quantized",
52
+ "weight_merge_groups": null
53
+ },
54
+ "global_quant_config": {
55
+ "bias": null,
56
+ "input_tensors": {
57
+ "ch_axis": -1,
58
+ "dtype": "fp4",
59
+ "group_size": 32,
60
+ "is_dynamic": true,
61
+ "is_scale_quant": false,
62
+ "mx_element_dtype": null,
63
+ "observer_cls": "PerBlockMXObserver",
64
+ "qscheme": "per_group",
65
+ "round_method": "half_even",
66
+ "scale_calculation_mode": "even",
67
+ "scale_format": "e8m0",
68
+ "scale_type": "float",
69
+ "symmetric": null
70
+ },
71
+ "output_tensors": null,
72
+ "target_device": null,
73
+ "weight": {
74
+ "ch_axis": -1,
75
+ "dtype": "fp4",
76
+ "group_size": 32,
77
+ "is_dynamic": false,
78
+ "is_scale_quant": false,
79
+ "mx_element_dtype": null,
80
+ "observer_cls": "PerBlockMXObserver",
81
+ "qscheme": "per_group",
82
+ "round_method": "half_even",
83
+ "scale_calculation_mode": "even",
84
+ "scale_format": "e8m0",
85
+ "scale_type": "float",
86
+ "symmetric": null
87
+ }
88
+ },
89
+ "kv_cache_quant_config": {
90
+ "*k_proj": {
91
+ "bias": null,
92
+ "input_tensors": {
93
+ "ch_axis": -1,
94
+ "dtype": "fp4",
95
+ "group_size": 32,
96
+ "is_dynamic": true,
97
+ "is_scale_quant": false,
98
+ "mx_element_dtype": null,
99
+ "observer_cls": "PerBlockMXObserver",
100
+ "qscheme": "per_group",
101
+ "round_method": "half_even",
102
+ "scale_calculation_mode": "even",
103
+ "scale_format": "e8m0",
104
+ "scale_type": "float",
105
+ "symmetric": null
106
+ },
107
+ "output_tensors": {
108
+ "ch_axis": null,
109
+ "dtype": "fp8_e4m3",
110
+ "group_size": null,
111
+ "is_dynamic": false,
112
+ "is_scale_quant": false,
113
+ "mx_element_dtype": null,
114
+ "observer_cls": "PerTensorMinMaxObserver",
115
+ "qscheme": "per_tensor",
116
+ "round_method": null,
117
+ "scale_calculation_mode": null,
118
+ "scale_format": null,
119
+ "scale_type": null,
120
+ "symmetric": null
121
+ },
122
+ "target_device": null,
123
+ "weight": {
124
+ "ch_axis": -1,
125
+ "dtype": "fp4",
126
+ "group_size": 32,
127
+ "is_dynamic": false,
128
+ "is_scale_quant": false,
129
+ "mx_element_dtype": null,
130
+ "observer_cls": "PerBlockMXObserver",
131
+ "qscheme": "per_group",
132
+ "round_method": "half_even",
133
+ "scale_calculation_mode": "even",
134
+ "scale_format": "e8m0",
135
+ "scale_type": "float",
136
+ "symmetric": null
137
+ }
138
+ },
139
+ "*v_proj": {
140
+ "bias": null,
141
+ "input_tensors": {
142
+ "ch_axis": -1,
143
+ "dtype": "fp4",
144
+ "group_size": 32,
145
+ "is_dynamic": true,
146
+ "is_scale_quant": false,
147
+ "mx_element_dtype": null,
148
+ "observer_cls": "PerBlockMXObserver",
149
+ "qscheme": "per_group",
150
+ "round_method": "half_even",
151
+ "scale_calculation_mode": "even",
152
+ "scale_format": "e8m0",
153
+ "scale_type": "float",
154
+ "symmetric": null
155
+ },
156
+ "output_tensors": {
157
+ "ch_axis": null,
158
+ "dtype": "fp8_e4m3",
159
+ "group_size": null,
160
+ "is_dynamic": false,
161
+ "is_scale_quant": false,
162
+ "mx_element_dtype": null,
163
+ "observer_cls": "PerTensorMinMaxObserver",
164
+ "qscheme": "per_tensor",
165
+ "round_method": null,
166
+ "scale_calculation_mode": null,
167
+ "scale_format": null,
168
+ "scale_type": null,
169
+ "symmetric": null
170
+ },
171
+ "target_device": null,
172
+ "weight": {
173
+ "ch_axis": -1,
174
+ "dtype": "fp4",
175
+ "group_size": 32,
176
+ "is_dynamic": false,
177
+ "is_scale_quant": false,
178
+ "mx_element_dtype": null,
179
+ "observer_cls": "PerBlockMXObserver",
180
+ "qscheme": "per_group",
181
+ "round_method": "half_even",
182
+ "scale_calculation_mode": "even",
183
+ "scale_format": "e8m0",
184
+ "scale_type": "float",
185
+ "symmetric": null
186
+ }
187
+ }
188
+ },
189
+ "layer_quant_config": {
190
+ "*k_proj": {
191
+ "bias": null,
192
+ "input_tensors": {
193
+ "ch_axis": -1,
194
+ "dtype": "fp4",
195
+ "group_size": 32,
196
+ "is_dynamic": true,
197
+ "is_scale_quant": false,
198
+ "mx_element_dtype": null,
199
+ "observer_cls": "PerBlockMXObserver",
200
+ "qscheme": "per_group",
201
+ "round_method": "half_even",
202
+ "scale_calculation_mode": "even",
203
+ "scale_format": "e8m0",
204
+ "scale_type": "float",
205
+ "symmetric": null
206
+ },
207
+ "output_tensors": {
208
+ "ch_axis": null,
209
+ "dtype": "fp8_e4m3",
210
+ "group_size": null,
211
+ "is_dynamic": false,
212
+ "is_scale_quant": false,
213
+ "mx_element_dtype": null,
214
+ "observer_cls": "PerTensorMinMaxObserver",
215
+ "qscheme": "per_tensor",
216
+ "round_method": null,
217
+ "scale_calculation_mode": null,
218
+ "scale_format": null,
219
+ "scale_type": null,
220
+ "symmetric": null
221
+ },
222
+ "target_device": null,
223
+ "weight": {
224
+ "ch_axis": -1,
225
+ "dtype": "fp4",
226
+ "group_size": 32,
227
+ "is_dynamic": false,
228
+ "is_scale_quant": false,
229
+ "mx_element_dtype": null,
230
+ "observer_cls": "PerBlockMXObserver",
231
+ "qscheme": "per_group",
232
+ "round_method": "half_even",
233
+ "scale_calculation_mode": "even",
234
+ "scale_format": "e8m0",
235
+ "scale_type": "float",
236
+ "symmetric": null
237
+ }
238
+ },
239
+ "*v_proj": {
240
+ "bias": null,
241
+ "input_tensors": {
242
+ "ch_axis": -1,
243
+ "dtype": "fp4",
244
+ "group_size": 32,
245
+ "is_dynamic": true,
246
+ "is_scale_quant": false,
247
+ "mx_element_dtype": null,
248
+ "observer_cls": "PerBlockMXObserver",
249
+ "qscheme": "per_group",
250
+ "round_method": "half_even",
251
+ "scale_calculation_mode": "even",
252
+ "scale_format": "e8m0",
253
+ "scale_type": "float",
254
+ "symmetric": null
255
+ },
256
+ "output_tensors": {
257
+ "ch_axis": null,
258
+ "dtype": "fp8_e4m3",
259
+ "group_size": null,
260
+ "is_dynamic": false,
261
+ "is_scale_quant": false,
262
+ "mx_element_dtype": null,
263
+ "observer_cls": "PerTensorMinMaxObserver",
264
+ "qscheme": "per_tensor",
265
+ "round_method": null,
266
+ "scale_calculation_mode": null,
267
+ "scale_format": null,
268
+ "scale_type": null,
269
+ "symmetric": null
270
+ },
271
+ "target_device": null,
272
+ "weight": {
273
+ "ch_axis": -1,
274
+ "dtype": "fp4",
275
+ "group_size": 32,
276
+ "is_dynamic": false,
277
+ "is_scale_quant": false,
278
+ "mx_element_dtype": null,
279
+ "observer_cls": "PerBlockMXObserver",
280
+ "qscheme": "per_group",
281
+ "round_method": "half_even",
282
+ "scale_calculation_mode": "even",
283
+ "scale_format": "e8m0",
284
+ "scale_type": "float",
285
+ "symmetric": null
286
+ }
287
+ }
288
+ },
289
+ "layer_type_quant_config": {},
290
+ "quant_method": "quark",
291
+ "quant_mode": "eager_mode",
292
+ "softmax_quant_spec": null,
293
+ "version": "0.10+d77d063292"
294
+ },
295
+ "rms_norm_eps": 1e-05,
296
+ "rope_scaling": null,
297
+ "rope_theta": 10000.0,
298
+ "tie_word_embeddings": false,
299
+ "torch_dtype": "float16",
300
+ "transformers_version": "4.51.0",
301
+ "use_cache": true,
302
+ "vocab_size": 32000
303
+ }
generation_config.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token_id": 1,
3
+ "do_sample": true,
4
+ "eos_token_id": 2,
5
+ "max_length": 4096,
6
+ "pad_token_id": 0,
7
+ "temperature": 0.6,
8
+ "top_p": 0.9,
9
+ "transformers_version": "4.51.0"
10
+ }
model-00001-of-00008.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1b8f732b8e2f332b01cfd0a40f3db86178692e622f017c9c35b0973001f69b6d
3
+ size 4945399768
model-00002-of-00008.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8adb2c200f5b8b9979b022b833ff9eea2ac15f02cd7df0f9acf7c8817a120732
3
+ size 4993177708
model-00003-of-00008.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b205513873c53abad82f0499f361f911f9c425d393f47aea1541d0134b51f050
3
+ size 4883077244
model-00004-of-00008.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0d13ba4ac3285da270e1cde2a52d4e38d24fbcaaaa2f25262fc8b51f7d97e4ba
3
+ size 4993177724
model-00005-of-00008.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:97b922f56e64b2e249220f6869b321f3fd1bb98d20b231238e6c1386b07c2763
3
+ size 4883077252
model-00006-of-00008.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:445162eb076e555be01a88351d01ee44651406dbf59aa8a527d944acd235b8cf
3
+ size 4998420604
model-00007-of-00008.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7aa59a11c8d7a0879eb342e5269d43dc77d236a5d1a8188a630311ed0a921f11
3
+ size 4966963324
model-00008-of-00008.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6579f58ce48c18a1bfbbb09a072114cd0ac7e9b23cb55721c1f6bed801cdb622
3
+ size 2752701968
model.safetensors.index.json ADDED
The diff for this file is too large to render. See raw diff
 
special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "</s>",
17
+ "unk_token": {
18
+ "content": "<unk>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
tokenizer_config.json ADDED
@@ -0,0 +1,45 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "add_prefix_space": true,
5
+ "added_tokens_decoder": {
6
+ "0": {
7
+ "content": "<unk>",
8
+ "lstrip": false,
9
+ "normalized": false,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "1": {
15
+ "content": "<s>",
16
+ "lstrip": false,
17
+ "normalized": false,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": true
21
+ },
22
+ "2": {
23
+ "content": "</s>",
24
+ "lstrip": false,
25
+ "normalized": false,
26
+ "rstrip": false,
27
+ "single_word": false,
28
+ "special": true
29
+ }
30
+ },
31
+ "bos_token": "<s>",
32
+ "chat_template": "{% if messages[0]['role'] == 'system' %}{% set loop_messages = messages[1:] %}{% set system_message = messages[0]['content'] %}{% else %}{% set loop_messages = messages %}{% set system_message = false %}{% endif %}{% for message in loop_messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if loop.index0 == 0 and system_message != false %}{% set content = '<<SYS>>\\n' + system_message + '\\n<</SYS>>\\n\\n' + message['content'] %}{% else %}{% set content = message['content'] %}{% endif %}{% if message['role'] == 'user' %}{{ bos_token + '[INST] ' + content.strip() + ' [/INST]' }}{% elif message['role'] == 'assistant' %}{{ ' ' + content.strip() + ' ' + eos_token }}{% endif %}{% endfor %}",
33
+ "clean_up_tokenization_spaces": false,
34
+ "eos_token": "</s>",
35
+ "extra_special_tokens": {},
36
+ "legacy": false,
37
+ "model_max_length": 1000000000000000019884624838656,
38
+ "pad_token": "</s>",
39
+ "padding_side": "left",
40
+ "sp_model_kwargs": {},
41
+ "spaces_between_special_tokens": false,
42
+ "tokenizer_class": "LlamaTokenizer",
43
+ "unk_token": "<unk>",
44
+ "use_default_system_prompt": false
45
+ }