Hugging Face's logo

aws-neuron
/
optimum-neuron-cache

Model card Files
xet
Community
611
dacorvo HF Staff commited on
Commit
01cf58b
·
verified ·
1 Parent(s): bea3550

Synchronizing local compiler cache.

Browse files
Files changed (11) hide show
  1. .gitattributes +3 -0
  2. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.3.dev2/llama4_text/meta-llama/Llama-4-Maverick-17B-128E-Instruct/efb739dd01c78aab1023.json +190 -0
  3. neuronxcc-2.21.33363.0+82129205/MODULE_0cd1e43f10774324aef0+ac10809c/compile_flags.json +1 -0
  4. neuronxcc-2.21.33363.0+82129205/MODULE_0cd1e43f10774324aef0+ac10809c/model.done +0 -0
  5. neuronxcc-2.21.33363.0+82129205/MODULE_0cd1e43f10774324aef0+ac10809c/model.hlo_module.pb +3 -0
  6. neuronxcc-2.21.33363.0+82129205/MODULE_0cd1e43f10774324aef0+ac10809c/model.neff +3 -0
  7. neuronxcc-2.21.33363.0+82129205/MODULE_0cd1e43f10774324aef0+ac10809c/wrapped_neff.hlo +3 -0
  8. neuronxcc-2.21.33363.0+82129205/MODULE_5fd694560d236ff1efdd+a32116a7/compile_flags.json +1 -0
  9. neuronxcc-2.21.33363.0+82129205/MODULE_5fd694560d236ff1efdd+a32116a7/model.done +0 -0
  10. neuronxcc-2.21.33363.0+82129205/MODULE_5fd694560d236ff1efdd+a32116a7/model.hlo_module.pb +3 -0
  11. neuronxcc-2.21.33363.0+82129205/MODULE_5fd694560d236ff1efdd+a32116a7/model.neff +3 -0
.gitattributes CHANGED
@@ -13766,3 +13766,6 @@ neuronxcc-2.21.33363.0+82129205/MODULE_d62ea26d65914ed758c0+a02c3a36/wrapped_nef
13766
  neuronxcc-2.21.33363.0+82129205/MODULE_8aac1344d1eb0e6a0a02+a02c3a36/model.neff filter=lfs diff=lfs merge=lfs -text
13767
  neuronxcc-2.21.33363.0+82129205/MODULE_8aac1344d1eb0e6a0a02+a02c3a36/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
13768
  neuronxcc-2.21.33363.0+82129205/MODULE_adc08173b01a5e1680e1+24129607/model.neff filter=lfs diff=lfs merge=lfs -text
 
 
 
 
13766
  neuronxcc-2.21.33363.0+82129205/MODULE_8aac1344d1eb0e6a0a02+a02c3a36/model.neff filter=lfs diff=lfs merge=lfs -text
13767
  neuronxcc-2.21.33363.0+82129205/MODULE_8aac1344d1eb0e6a0a02+a02c3a36/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
13768
  neuronxcc-2.21.33363.0+82129205/MODULE_adc08173b01a5e1680e1+24129607/model.neff filter=lfs diff=lfs merge=lfs -text
13769
+ neuronxcc-2.21.33363.0+82129205/MODULE_0cd1e43f10774324aef0+ac10809c/model.neff filter=lfs diff=lfs merge=lfs -text
13770
+ neuronxcc-2.21.33363.0+82129205/MODULE_0cd1e43f10774324aef0+ac10809c/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
13771
+ neuronxcc-2.21.33363.0+82129205/MODULE_5fd694560d236ff1efdd+a32116a7/model.neff filter=lfs diff=lfs merge=lfs -text
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.3.dev2/llama4_text/meta-llama/Llama-4-Maverick-17B-128E-Instruct/efb739dd01c78aab1023.json ADDED
@@ -0,0 +1,190 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "meta-llama/Llama-4-Maverick-17B-128E-Instruct",
4
+ "_task": "text-generation",
5
+ "attention_bias": false,
6
+ "attention_chunk_size": 8192,
7
+ "attention_dropout": 0.0,
8
+ "attn_scale": 0.1,
9
+ "attn_temperature_tuning": true,
10
+ "dtype": "bfloat16",
11
+ "floor_scale": 8192,
12
+ "for_llm_compressor": false,
13
+ "head_dim": 128,
14
+ "hidden_act": "silu",
15
+ "hidden_size": 5120,
16
+ "initializer_range": 0.02,
17
+ "interleave_moe_layer_step": 2,
18
+ "intermediate_size": 8192,
19
+ "intermediate_size_mlp": 16384,
20
+ "layer_types": [
21
+ "chunked_attention",
22
+ "chunked_attention",
23
+ "chunked_attention",
24
+ "full_attention",
25
+ "chunked_attention",
26
+ "chunked_attention",
27
+ "chunked_attention",
28
+ "full_attention",
29
+ "chunked_attention",
30
+ "chunked_attention",
31
+ "chunked_attention",
32
+ "full_attention",
33
+ "chunked_attention",
34
+ "chunked_attention",
35
+ "chunked_attention",
36
+ "full_attention",
37
+ "chunked_attention",
38
+ "chunked_attention",
39
+ "chunked_attention",
40
+ "full_attention",
41
+ "chunked_attention",
42
+ "chunked_attention",
43
+ "chunked_attention",
44
+ "full_attention",
45
+ "chunked_attention",
46
+ "chunked_attention",
47
+ "chunked_attention",
48
+ "full_attention",
49
+ "chunked_attention",
50
+ "chunked_attention",
51
+ "chunked_attention",
52
+ "full_attention",
53
+ "chunked_attention",
54
+ "chunked_attention",
55
+ "chunked_attention",
56
+ "full_attention",
57
+ "chunked_attention",
58
+ "chunked_attention",
59
+ "chunked_attention",
60
+ "full_attention",
61
+ "chunked_attention",
62
+ "chunked_attention",
63
+ "chunked_attention",
64
+ "full_attention",
65
+ "chunked_attention",
66
+ "chunked_attention",
67
+ "chunked_attention",
68
+ "full_attention"
69
+ ],
70
+ "max_position_embeddings": 1048576,
71
+ "model_type": "llama4_text",
72
+ "moe_layers": [
73
+ 1,
74
+ 3,
75
+ 5,
76
+ 7,
77
+ 9,
78
+ 11,
79
+ 13,
80
+ 15,
81
+ 17,
82
+ 19,
83
+ 21,
84
+ 23,
85
+ 25,
86
+ 27,
87
+ 29,
88
+ 31,
89
+ 33,
90
+ 35,
91
+ 37,
92
+ 39,
93
+ 41,
94
+ 43,
95
+ 45,
96
+ 47
97
+ ],
98
+ "neuron": {
99
+ "_serialized_key": "NxDNeuronConfig",
100
+ "batch_size": 1,
101
+ "capacity_factor": null,
102
+ "checkpoint_id": "meta-llama/Llama-4-Maverick-17B-128E-Instruct",
103
+ "checkpoint_revision": "73d14711bcc77c16df3470856949c3764056b617",
104
+ "continuous_batching": false,
105
+ "ep_degree": 1,
106
+ "fused_qkv": false,
107
+ "glu_mlp": true,
108
+ "local_ranks_size": 64,
109
+ "max_batch_size": 1,
110
+ "max_context_length": 4096,
111
+ "max_topk": 256,
112
+ "n_active_tokens": 4096,
113
+ "neuronxcc_version": "2.21.33363.0+82129205",
114
+ "on_device_sampling": true,
115
+ "optimum_neuron_version": "0.4.3.dev2",
116
+ "output_logits": false,
117
+ "pp_degree": 1,
118
+ "sequence_length": 4096,
119
+ "speculation_length": 0,
120
+ "start_rank_id": 0,
121
+ "target": "trn2",
122
+ "torch_dtype": "bfloat16",
123
+ "tp_degree": 64
124
+ },
125
+ "no_rope_layers": [
126
+ 1,
127
+ 1,
128
+ 1,
129
+ 0,
130
+ 1,
131
+ 1,
132
+ 1,
133
+ 0,
134
+ 1,
135
+ 1,
136
+ 1,
137
+ 0,
138
+ 1,
139
+ 1,
140
+ 1,
141
+ 0,
142
+ 1,
143
+ 1,
144
+ 1,
145
+ 0,
146
+ 1,
147
+ 1,
148
+ 1,
149
+ 0,
150
+ 1,
151
+ 1,
152
+ 1,
153
+ 0,
154
+ 1,
155
+ 1,
156
+ 1,
157
+ 0,
158
+ 1,
159
+ 1,
160
+ 1,
161
+ 0,
162
+ 1,
163
+ 1,
164
+ 1,
165
+ 0,
166
+ 1,
167
+ 1,
168
+ 1,
169
+ 0,
170
+ 1,
171
+ 1,
172
+ 1,
173
+ 0
174
+ ],
175
+ "num_attention_heads": 40,
176
+ "num_experts_per_tok": 1,
177
+ "num_hidden_layers": 48,
178
+ "num_key_value_heads": 8,
179
+ "num_local_experts": 128,
180
+ "output_router_logits": false,
181
+ "rms_norm_eps": 1e-05,
182
+ "rope_scaling": null,
183
+ "rope_theta": 500000.0,
184
+ "router_aux_loss_coef": 0.001,
185
+ "router_jitter_noise": 0.0,
186
+ "tie_word_embeddings": false,
187
+ "use_cache": true,
188
+ "use_qk_norm": false,
189
+ "vocab_size": 202048
190
+ }
neuronxcc-2.21.33363.0+82129205/MODULE_0cd1e43f10774324aef0+ac10809c/compile_flags.json ADDED
@@ -0,0 +1 @@
 
 
1
+ ["--target=trn2", "--auto-cast=none", "--model-type=transformer", "--tensorizer-options=--enable-ccop-compute-overlap --cc-pipeline-tiling-factor=2 --vectorize-strided-dma ", "-O2", "--lnc=2", "--logfile=/tmp/nxd_model/token_generation/_tp0_bk0/log-neuron-cc.txt", "--enable-internal-neff-wrapper"]
neuronxcc-2.21.33363.0+82129205/MODULE_0cd1e43f10774324aef0+ac10809c/model.done ADDED
File without changes
neuronxcc-2.21.33363.0+82129205/MODULE_0cd1e43f10774324aef0+ac10809c/model.hlo_module.pb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:92d49d02bddfa3637b9147a4c4f23494366e66e76b339a2ab0df4fbf0ea57b4c
3
+ size 103022001
neuronxcc-2.21.33363.0+82129205/MODULE_0cd1e43f10774324aef0+ac10809c/model.neff ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0374fac8dd8c5408cf266c7f2619c99f5bff7bc383f0f543e420184d87d5a406
3
+ size 7005184
neuronxcc-2.21.33363.0+82129205/MODULE_0cd1e43f10774324aef0+ac10809c/wrapped_neff.hlo ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bd5fefe567d2b0cce5d94b1b68e6a67fcfbd1d3f0c0be306061611873a1b3dfb
3
+ size 7314549
neuronxcc-2.21.33363.0+82129205/MODULE_5fd694560d236ff1efdd+a32116a7/compile_flags.json ADDED
@@ -0,0 +1 @@
 
 
1
+ ["--target=trn2", "--auto-cast=none", "--model-type=transformer", "--tensorizer-options=--enable-ccop-compute-overlap --cc-pipeline-tiling-factor=2 --vectorize-strided-dma ", "-O2", "--lnc=2", "--logfile=/tmp/nxd_model/context_encoding/_tp0_bk0/log-neuron-cc.txt"]
neuronxcc-2.21.33363.0+82129205/MODULE_5fd694560d236ff1efdd+a32116a7/model.done ADDED
File without changes
neuronxcc-2.21.33363.0+82129205/MODULE_5fd694560d236ff1efdd+a32116a7/model.hlo_module.pb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:172361ce98f10e17aa79bec0fb0110b335cc4dca6597945e3748dbf3e55001fe
3
+ size 107209277
neuronxcc-2.21.33363.0+82129205/MODULE_5fd694560d236ff1efdd+a32116a7/model.neff ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c0ecfa8396e097de6a816190f3f948720db16ed4bee4af252431274a86e3bd03
3
+ size 39906304