Hugging Face's logo

aws-neuron
/
optimum-neuron-cache

Model card Files
xet
 Community
709
dacorvo HF Staff commited on
Commit
a96bb4f
·
verified ·
1 Parent(s): 10f071c

Synchronizing local compiler cache.

Browse files
Files changed (12) hide show
  1. .gitattributes +3 -0
  2. neuronxcc-2.26.6360.0+6f180f47/0_REGISTRY/0.4.7.dev0/fe4aef5e4c88bcaa8eae64c5f71d7b0c1a5e475bbfde7288f59fbde13a1c8732/cd301a857ca92f6646f1.json +191 -0
  3. neuronxcc-2.26.6360.0+6f180f47/0_REGISTRY/0.4.7.dev0/llama4_text/meta-llama/Llama-4-Maverick-17B-128E-Instruct/cd301a857ca92f6646f1.json +191 -0
  4. neuronxcc-2.26.6360.0+6f180f47/MODULE_85b1b84a9d438b83ec01+4a27301b/compile_flags.json +1 -0
  5. neuronxcc-2.26.6360.0+6f180f47/MODULE_85b1b84a9d438b83ec01+4a27301b/model.done +0 -0
  6. neuronxcc-2.26.6360.0+6f180f47/MODULE_85b1b84a9d438b83ec01+4a27301b/model.hlo_module.pb +3 -0
  7. neuronxcc-2.26.6360.0+6f180f47/MODULE_85b1b84a9d438b83ec01+4a27301b/model.neff +3 -0
  8. neuronxcc-2.26.6360.0+6f180f47/MODULE_85b1b84a9d438b83ec01+4a27301b/wrapped_neff.hlo +3 -0
  9. neuronxcc-2.26.6360.0+6f180f47/MODULE_8958f1ff32ed1da354e9+665810b0/compile_flags.json +1 -0
  10. neuronxcc-2.26.6360.0+6f180f47/MODULE_8958f1ff32ed1da354e9+665810b0/model.done +0 -0
  11. neuronxcc-2.26.6360.0+6f180f47/MODULE_8958f1ff32ed1da354e9+665810b0/model.hlo_module.pb +3 -0
  12. neuronxcc-2.26.6360.0+6f180f47/MODULE_8958f1ff32ed1da354e9+665810b0/model.neff +3 -0
.gitattributes CHANGED
@@ -18161,3 +18161,6 @@ neuronxcc-2.26.6360.0+6f180f47/MODULE_d3c738df0d70fb71aee1+12dd9119/model.neff f
18161
  neuronxcc-2.26.6360.0+6f180f47/MODULE_94a18ee40f5193ca700f+84337dd9/model.neff filter=lfs diff=lfs merge=lfs -text
18162
  neuronxcc-2.26.6360.0+6f180f47/MODULE_94a18ee40f5193ca700f+84337dd9/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
18163
  neuronxcc-2.26.6360.0+6f180f47/MODULE_ac79315975f811c8fe86+12dd9119/model.neff filter=lfs diff=lfs merge=lfs -text
 
 
 
 
18161
  neuronxcc-2.26.6360.0+6f180f47/MODULE_94a18ee40f5193ca700f+84337dd9/model.neff filter=lfs diff=lfs merge=lfs -text
18162
  neuronxcc-2.26.6360.0+6f180f47/MODULE_94a18ee40f5193ca700f+84337dd9/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
18163
  neuronxcc-2.26.6360.0+6f180f47/MODULE_ac79315975f811c8fe86+12dd9119/model.neff filter=lfs diff=lfs merge=lfs -text
18164
+ neuronxcc-2.26.6360.0+6f180f47/MODULE_85b1b84a9d438b83ec01+4a27301b/model.neff filter=lfs diff=lfs merge=lfs -text
18165
+ neuronxcc-2.26.6360.0+6f180f47/MODULE_85b1b84a9d438b83ec01+4a27301b/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
18166
+ neuronxcc-2.26.6360.0+6f180f47/MODULE_8958f1ff32ed1da354e9+665810b0/model.neff filter=lfs diff=lfs merge=lfs -text
neuronxcc-2.26.6360.0+6f180f47/0_REGISTRY/0.4.7.dev0/fe4aef5e4c88bcaa8eae64c5f71d7b0c1a5e475bbfde7288f59fbde13a1c8732/cd301a857ca92f6646f1.json ADDED
@@ -0,0 +1,191 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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": "2.26.6360.0+6f180f47",
114
+ "on_device_sampling": true,
115
+ "optimum_neuron_version": "0.4.7.dev0",
116
+ "output_logits": false,
117
+ "pp_degree": 1,
118
+ "prefill_chunk_size": 1024,
119
+ "sequence_length": 4096,
120
+ "speculation_length": 0,
121
+ "start_rank_id": 0,
122
+ "target": "trn2",
123
+ "torch_dtype": "bfloat16",
124
+ "tp_degree": 64
125
+ },
126
+ "no_rope_layers": [
127
+ 1,
128
+ 1,
129
+ 1,
130
+ 0,
131
+ 1,
132
+ 1,
133
+ 1,
134
+ 0,
135
+ 1,
136
+ 1,
137
+ 1,
138
+ 0,
139
+ 1,
140
+ 1,
141
+ 1,
142
+ 0,
143
+ 1,
144
+ 1,
145
+ 1,
146
+ 0,
147
+ 1,
148
+ 1,
149
+ 1,
150
+ 0,
151
+ 1,
152
+ 1,
153
+ 1,
154
+ 0,
155
+ 1,
156
+ 1,
157
+ 1,
158
+ 0,
159
+ 1,
160
+ 1,
161
+ 1,
162
+ 0,
163
+ 1,
164
+ 1,
165
+ 1,
166
+ 0,
167
+ 1,
168
+ 1,
169
+ 1,
170
+ 0,
171
+ 1,
172
+ 1,
173
+ 1,
174
+ 0
175
+ ],
176
+ "num_attention_heads": 40,
177
+ "num_experts_per_tok": 1,
178
+ "num_hidden_layers": 48,
179
+ "num_key_value_heads": 8,
180
+ "num_local_experts": 128,
181
+ "output_router_logits": false,
182
+ "rms_norm_eps": 1e-05,
183
+ "rope_scaling": null,
184
+ "rope_theta": 500000.0,
185
+ "router_aux_loss_coef": 0.001,
186
+ "router_jitter_noise": 0.0,
187
+ "tie_word_embeddings": false,
188
+ "use_cache": true,
189
+ "use_qk_norm": false,
190
+ "vocab_size": 202048
191
+ }
neuronxcc-2.26.6360.0+6f180f47/0_REGISTRY/0.4.7.dev0/llama4_text/meta-llama/Llama-4-Maverick-17B-128E-Instruct/cd301a857ca92f6646f1.json ADDED
@@ -0,0 +1,191 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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": "2.26.6360.0+6f180f47",
114
+ "on_device_sampling": true,
115
+ "optimum_neuron_version": "0.4.7.dev0",
116
+ "output_logits": false,
117
+ "pp_degree": 1,
118
+ "prefill_chunk_size": 1024,
119
+ "sequence_length": 4096,
120
+ "speculation_length": 0,
121
+ "start_rank_id": 0,
122
+ "target": "trn2",
123
+ "torch_dtype": "bfloat16",
124
+ "tp_degree": 64
125
+ },
126
+ "no_rope_layers": [
127
+ 1,
128
+ 1,
129
+ 1,
130
+ 0,
131
+ 1,
132
+ 1,
133
+ 1,
134
+ 0,
135
+ 1,
136
+ 1,
137
+ 1,
138
+ 0,
139
+ 1,
140
+ 1,
141
+ 1,
142
+ 0,
143
+ 1,
144
+ 1,
145
+ 1,
146
+ 0,
147
+ 1,
148
+ 1,
149
+ 1,
150
+ 0,
151
+ 1,
152
+ 1,
153
+ 1,
154
+ 0,
155
+ 1,
156
+ 1,
157
+ 1,
158
+ 0,
159
+ 1,
160
+ 1,
161
+ 1,
162
+ 0,
163
+ 1,
164
+ 1,
165
+ 1,
166
+ 0,
167
+ 1,
168
+ 1,
169
+ 1,
170
+ 0,
171
+ 1,
172
+ 1,
173
+ 1,
174
+ 0
175
+ ],
176
+ "num_attention_heads": 40,
177
+ "num_experts_per_tok": 1,
178
+ "num_hidden_layers": 48,
179
+ "num_key_value_heads": 8,
180
+ "num_local_experts": 128,
181
+ "output_router_logits": false,
182
+ "rms_norm_eps": 1e-05,
183
+ "rope_scaling": null,
184
+ "rope_theta": 500000.0,
185
+ "router_aux_loss_coef": 0.001,
186
+ "router_jitter_noise": 0.0,
187
+ "tie_word_embeddings": false,
188
+ "use_cache": true,
189
+ "use_qk_norm": false,
190
+ "vocab_size": 202048
191
+ }
neuronxcc-2.26.6360.0+6f180f47/MODULE_85b1b84a9d438b83ec01+4a27301b/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", "--verbose=35", "--logfile=/tmp/nxd_model/token_generation/_tp0_bk0/log-neuron-cc.txt", "--enable-internal-neff-wrapper"]
neuronxcc-2.26.6360.0+6f180f47/MODULE_85b1b84a9d438b83ec01+4a27301b/model.done ADDED
File without changes
neuronxcc-2.26.6360.0+6f180f47/MODULE_85b1b84a9d438b83ec01+4a27301b/model.hlo_module.pb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f82ff652c70525bb744c3aaf18c40182eabc6a946364684ded177f37181219da
3
+ size 106624139
neuronxcc-2.26.6360.0+6f180f47/MODULE_85b1b84a9d438b83ec01+4a27301b/model.neff ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d7b1710e789f523c9b4a3fc4ffa91271396eab109900f83794b5a0c9a6e5187e
3
+ size 6284235
neuronxcc-2.26.6360.0+6f180f47/MODULE_85b1b84a9d438b83ec01+4a27301b/wrapped_neff.hlo ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:65f5e8a1d5fd08371d999d2a82ea82465427bd86408534210dc05e4506fc0851
3
+ size 6593591
neuronxcc-2.26.6360.0+6f180f47/MODULE_8958f1ff32ed1da354e9+665810b0/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", "--verbose=35", "--logfile=/tmp/nxd_model/chunked_prefill/_tp0_bk0/log-neuron-cc.txt"]
neuronxcc-2.26.6360.0+6f180f47/MODULE_8958f1ff32ed1da354e9+665810b0/model.done ADDED
File without changes
neuronxcc-2.26.6360.0+6f180f47/MODULE_8958f1ff32ed1da354e9+665810b0/model.hlo_module.pb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:913f1ffbd6ba199413eb65b8904213f095cdaf3b3187f55d7698c02362cff289
3
+ size 126976151
neuronxcc-2.26.6360.0+6f180f47/MODULE_8958f1ff32ed1da354e9+665810b0/model.neff ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:543c294ce2d888a450be41a71eadd939ee8c0942fb2b61933d9bf74a2c41372e
3
+ size 24566566