Hugging Face's logo

optimum-internal-testing
/
neuron-testing-cache

Model card Files
xet
 Community
dacorvo HF Staff commited on
Commit
75f5ba3
·
verified ·
1 Parent(s): 93886a6

Synchronizing local compiler cache.

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. .gitattributes +11 -0
  2. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.6.dev2/e0b6d1e2424243dcd9ff1755e02969dcc312d14df531d876c5c2892f285b2863/4e1fa9712696dd29a0bc.json +95 -0
  3. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.6.dev2/e0b6d1e2424243dcd9ff1755e02969dcc312d14df531d876c5c2892f285b2863/80514be9a02fddc9c476.json +95 -0
  4. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.6.dev2/qwen3/Qwen/Qwen3-Embedding-4B/80514be9a02fddc9c476.json +95 -0
  5. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.6.dev3/929b02754a13cbfdf657d863c3fc6f3bce672879bc6ae48ab45be21e881e9ec2/96baef0f2a01e5e29193.json +88 -0
  6. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.6.dev3/c54cbc73d9a74e547bf7ca1feb2b290b641ed261e32f7c08baba5633884f1298/d73c51dc7dd75010abe9.json +63 -0
  7. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.6.dev3/cf6b9a360dcf294104671106bae2adbd9fd291823bb60a351883163684073231/d182199b5d53c98ec562.json +64 -0
  8. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.6.dev3/e0b6d1e2424243dcd9ff1755e02969dcc312d14df531d876c5c2892f285b2863/24e7f0a205508b46b0eb.json +96 -0
  9. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.6.dev3/e0b6d1e2424243dcd9ff1755e02969dcc312d14df531d876c5c2892f285b2863/6e2489df1bc1e5c5af63.json +96 -0
  10. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.6.dev3/e0b6d1e2424243dcd9ff1755e02969dcc312d14df531d876c5c2892f285b2863/a34b748f9038b1e376dc.json +96 -0
  11. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.6.dev3/llama/meta-llama/Meta-Llama-3.1-8B-Instruct/d73c51dc7dd75010abe9.json +63 -0
  12. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.6.dev3/qwen3/Qwen/Qwen3-Embedding-4B/24e7f0a205508b46b0eb.json +96 -0
  13. neuronxcc-2.21.33363.0+82129205/MODULE_0352823045bfbd34fd3d+fb4cc044/compile_flags.json +1 -0
  14. neuronxcc-2.21.33363.0+82129205/MODULE_0352823045bfbd34fd3d+fb4cc044/model.done +0 -0
  15. neuronxcc-2.21.33363.0+82129205/MODULE_0352823045bfbd34fd3d+fb4cc044/model.hlo_module.pb +3 -0
  16. neuronxcc-2.21.33363.0+82129205/MODULE_0352823045bfbd34fd3d+fb4cc044/model.neff +3 -0
  17. neuronxcc-2.21.33363.0+82129205/MODULE_04dc6de9513dd8bb05c4+fb4cc044/compile_flags.json +1 -0
  18. neuronxcc-2.21.33363.0+82129205/MODULE_04dc6de9513dd8bb05c4+fb4cc044/model.done +0 -0
  19. neuronxcc-2.21.33363.0+82129205/MODULE_04dc6de9513dd8bb05c4+fb4cc044/model.hlo_module.pb +3 -0
  20. neuronxcc-2.21.33363.0+82129205/MODULE_04dc6de9513dd8bb05c4+fb4cc044/model.neff +3 -0
  21. neuronxcc-2.21.33363.0+82129205/MODULE_1759fced9dd9a678973e+24129607/compile_flags.json +1 -0
  22. neuronxcc-2.21.33363.0+82129205/MODULE_1759fced9dd9a678973e+24129607/model.done +0 -0
  23. neuronxcc-2.21.33363.0+82129205/MODULE_1759fced9dd9a678973e+24129607/model.hlo_module.pb +3 -0
  24. neuronxcc-2.21.33363.0+82129205/MODULE_1759fced9dd9a678973e+24129607/model.neff +3 -0
  25. neuronxcc-2.21.33363.0+82129205/MODULE_1824b665798bf04e2aa2+24129607/compile_flags.json +1 -0
  26. neuronxcc-2.21.33363.0+82129205/MODULE_1824b665798bf04e2aa2+24129607/model.done +0 -0
  27. neuronxcc-2.21.33363.0+82129205/MODULE_1824b665798bf04e2aa2+24129607/model.hlo_module.pb +3 -0
  28. neuronxcc-2.21.33363.0+82129205/MODULE_1824b665798bf04e2aa2+24129607/model.neff +3 -0
  29. neuronxcc-2.21.33363.0+82129205/MODULE_24a7a312e7cf7a6b6c5d+fb4cc044/compile_flags.json +1 -0
  30. neuronxcc-2.21.33363.0+82129205/MODULE_24a7a312e7cf7a6b6c5d+fb4cc044/model.done +0 -0
  31. neuronxcc-2.21.33363.0+82129205/MODULE_24a7a312e7cf7a6b6c5d+fb4cc044/model.hlo_module.pb +3 -0
  32. neuronxcc-2.21.33363.0+82129205/MODULE_24a7a312e7cf7a6b6c5d+fb4cc044/model.neff +3 -0
  33. neuronxcc-2.21.33363.0+82129205/MODULE_68b1b688bc907867445e+a02c3a36/model.hlo_module.pb +1 -1
  34. neuronxcc-2.21.33363.0+82129205/MODULE_68b1b688bc907867445e+a02c3a36/model.neff +1 -1
  35. neuronxcc-2.21.33363.0+82129205/MODULE_68b1b688bc907867445e+a02c3a36/wrapped_neff.hlo +1 -1
  36. neuronxcc-2.21.33363.0+82129205/MODULE_88833531cb6c582d4830+a02c3a36/compile_flags.json +1 -0
  37. neuronxcc-2.21.33363.0+82129205/MODULE_88833531cb6c582d4830+a02c3a36/model.done +0 -0
  38. neuronxcc-2.21.33363.0+82129205/MODULE_88833531cb6c582d4830+a02c3a36/model.hlo_module.pb +3 -0
  39. neuronxcc-2.21.33363.0+82129205/MODULE_88833531cb6c582d4830+a02c3a36/model.neff +3 -0
  40. neuronxcc-2.21.33363.0+82129205/MODULE_88833531cb6c582d4830+a02c3a36/wrapped_neff.hlo +3 -0
  41. neuronxcc-2.21.33363.0+82129205/MODULE_9068b179aef4ffa33207+6c043b9a/compile_flags.json +1 -0
  42. neuronxcc-2.21.33363.0+82129205/MODULE_9068b179aef4ffa33207+6c043b9a/model.hlo_module.pb +3 -0
  43. neuronxcc-2.21.33363.0+82129205/MODULE_92b96f10a3841c3c7919+6c043b9a/compile_flags.json +1 -0
  44. neuronxcc-2.21.33363.0+82129205/MODULE_92b96f10a3841c3c7919+6c043b9a/model.done +0 -0
  45. neuronxcc-2.21.33363.0+82129205/MODULE_92b96f10a3841c3c7919+6c043b9a/model.hlo_module.pb +3 -0
  46. neuronxcc-2.21.33363.0+82129205/MODULE_92b96f10a3841c3c7919+6c043b9a/model.neff +3 -0
  47. neuronxcc-2.21.33363.0+82129205/MODULE_95b9072f246645f24461+a02c3a36/model.hlo_module.pb +1 -1
  48. neuronxcc-2.21.33363.0+82129205/MODULE_95b9072f246645f24461+a02c3a36/model.neff +1 -1
  49. neuronxcc-2.21.33363.0+82129205/MODULE_95b9072f246645f24461+a02c3a36/wrapped_neff.hlo +1 -1
  50. neuronxcc-2.21.33363.0+82129205/MODULE_a431a415019174ce0469+6c043b9a/compile_flags.json +1 -0
.gitattributes CHANGED
@@ -7234,3 +7234,14 @@ neuronxcc-2.21.33363.0+82129205/MODULE_1662e89e76097dde374a+24129607/model.neff
7234
  neuronxcc-2.21.33363.0+82129205/MODULE_789b4718541dd2474859+a02c3a36/model.neff filter=lfs diff=lfs merge=lfs -text
7235
  neuronxcc-2.21.33363.0+82129205/MODULE_789b4718541dd2474859+a02c3a36/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
7236
  neuronxcc-2.21.33363.0+82129205/MODULE_ff54f59684fbb72ef7e9+6170d8e1/model.neff filter=lfs diff=lfs merge=lfs -text
 
 
 
 
 
 
 
 
 
 
 
 
7234
  neuronxcc-2.21.33363.0+82129205/MODULE_789b4718541dd2474859+a02c3a36/model.neff filter=lfs diff=lfs merge=lfs -text
7235
  neuronxcc-2.21.33363.0+82129205/MODULE_789b4718541dd2474859+a02c3a36/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
7236
  neuronxcc-2.21.33363.0+82129205/MODULE_ff54f59684fbb72ef7e9+6170d8e1/model.neff filter=lfs diff=lfs merge=lfs -text
7237
+ neuronxcc-2.21.33363.0+82129205/MODULE_0352823045bfbd34fd3d+fb4cc044/model.neff filter=lfs diff=lfs merge=lfs -text
7238
+ neuronxcc-2.21.33363.0+82129205/MODULE_04dc6de9513dd8bb05c4+fb4cc044/model.neff filter=lfs diff=lfs merge=lfs -text
7239
+ neuronxcc-2.21.33363.0+82129205/MODULE_1759fced9dd9a678973e+24129607/model.neff filter=lfs diff=lfs merge=lfs -text
7240
+ neuronxcc-2.21.33363.0+82129205/MODULE_1824b665798bf04e2aa2+24129607/model.neff filter=lfs diff=lfs merge=lfs -text
7241
+ neuronxcc-2.21.33363.0+82129205/MODULE_24a7a312e7cf7a6b6c5d+fb4cc044/model.neff filter=lfs diff=lfs merge=lfs -text
7242
+ neuronxcc-2.21.33363.0+82129205/MODULE_88833531cb6c582d4830+a02c3a36/model.neff filter=lfs diff=lfs merge=lfs -text
7243
+ neuronxcc-2.21.33363.0+82129205/MODULE_88833531cb6c582d4830+a02c3a36/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
7244
+ neuronxcc-2.21.33363.0+82129205/MODULE_92b96f10a3841c3c7919+6c043b9a/model.neff filter=lfs diff=lfs merge=lfs -text
7245
+ neuronxcc-2.21.33363.0+82129205/MODULE_a431a415019174ce0469+6c043b9a/model.neff filter=lfs diff=lfs merge=lfs -text
7246
+ neuronxcc-2.21.33363.0+82129205/MODULE_bbee1996ee8025809e33+a02c3a36/model.neff filter=lfs diff=lfs merge=lfs -text
7247
+ neuronxcc-2.21.33363.0+82129205/MODULE_bbee1996ee8025809e33+a02c3a36/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.6.dev2/e0b6d1e2424243dcd9ff1755e02969dcc312d14df531d876c5c2892f285b2863/4e1fa9712696dd29a0bc.json ADDED
@@ -0,0 +1,95 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "Qwen/Qwen3-Embedding-4B",
4
+ "_task": "feature-extraction",
5
+ "architectures": [
6
+ "Qwen3ForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "dtype": "bfloat16",
11
+ "head_dim": 128,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 2560,
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 9728,
16
+ "layer_types": [
17
+ "full_attention",
18
+ "full_attention",
19
+ "full_attention",
20
+ "full_attention",
21
+ "full_attention",
22
+ "full_attention",
23
+ "full_attention",
24
+ "full_attention",
25
+ "full_attention",
26
+ "full_attention",
27
+ "full_attention",
28
+ "full_attention",
29
+ "full_attention",
30
+ "full_attention",
31
+ "full_attention",
32
+ "full_attention",
33
+ "full_attention",
34
+ "full_attention",
35
+ "full_attention",
36
+ "full_attention",
37
+ "full_attention",
38
+ "full_attention",
39
+ "full_attention",
40
+ "full_attention",
41
+ "full_attention",
42
+ "full_attention",
43
+ "full_attention",
44
+ "full_attention",
45
+ "full_attention",
46
+ "full_attention",
47
+ "full_attention",
48
+ "full_attention",
49
+ "full_attention",
50
+ "full_attention",
51
+ "full_attention",
52
+ "full_attention"
53
+ ],
54
+ "max_position_embeddings": 40960,
55
+ "max_window_layers": 36,
56
+ "model_type": "qwen3",
57
+ "neuron": {
58
+ "_serialized_key": "NxDNeuronConfig",
59
+ "batch_size": 1,
60
+ "capacity_factor": null,
61
+ "checkpoint_id": "Qwen/Qwen3-Embedding-4B",
62
+ "checkpoint_revision": "5cf2132abc99cad020ac570b19d031efec650f2b",
63
+ "continuous_batching": false,
64
+ "ep_degree": 1,
65
+ "fused_qkv": true,
66
+ "glu_mlp": true,
67
+ "local_ranks_size": 2,
68
+ "max_batch_size": 1,
69
+ "max_context_length": 8192,
70
+ "max_topk": 256,
71
+ "n_active_tokens": 8192,
72
+ "neuronxcc_version": "2.21.33363.0+82129205",
73
+ "on_device_sampling": false,
74
+ "optimum_neuron_version": "0.4.6.dev2",
75
+ "output_logits": false,
76
+ "pp_degree": 1,
77
+ "sequence_length": 8192,
78
+ "speculation_length": 0,
79
+ "start_rank_id": 0,
80
+ "target": "trn1",
81
+ "torch_dtype": "bfloat16",
82
+ "tp_degree": 2
83
+ },
84
+ "num_attention_heads": 32,
85
+ "num_hidden_layers": 36,
86
+ "num_key_value_heads": 8,
87
+ "rms_norm_eps": 1e-06,
88
+ "rope_scaling": null,
89
+ "rope_theta": 1000000,
90
+ "sliding_window": null,
91
+ "tie_word_embeddings": true,
92
+ "use_cache": true,
93
+ "use_sliding_window": false,
94
+ "vocab_size": 151665
95
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.6.dev2/e0b6d1e2424243dcd9ff1755e02969dcc312d14df531d876c5c2892f285b2863/80514be9a02fddc9c476.json ADDED
@@ -0,0 +1,95 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "Qwen/Qwen3-Embedding-4B",
4
+ "_task": "feature-extraction",
5
+ "architectures": [
6
+ "Qwen3ForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "dtype": "bfloat16",
11
+ "head_dim": 128,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 2560,
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 9728,
16
+ "layer_types": [
17
+ "full_attention",
18
+ "full_attention",
19
+ "full_attention",
20
+ "full_attention",
21
+ "full_attention",
22
+ "full_attention",
23
+ "full_attention",
24
+ "full_attention",
25
+ "full_attention",
26
+ "full_attention",
27
+ "full_attention",
28
+ "full_attention",
29
+ "full_attention",
30
+ "full_attention",
31
+ "full_attention",
32
+ "full_attention",
33
+ "full_attention",
34
+ "full_attention",
35
+ "full_attention",
36
+ "full_attention",
37
+ "full_attention",
38
+ "full_attention",
39
+ "full_attention",
40
+ "full_attention",
41
+ "full_attention",
42
+ "full_attention",
43
+ "full_attention",
44
+ "full_attention",
45
+ "full_attention",
46
+ "full_attention",
47
+ "full_attention",
48
+ "full_attention",
49
+ "full_attention",
50
+ "full_attention",
51
+ "full_attention",
52
+ "full_attention"
53
+ ],
54
+ "max_position_embeddings": 40960,
55
+ "max_window_layers": 36,
56
+ "model_type": "qwen3",
57
+ "neuron": {
58
+ "_serialized_key": "NxDNeuronConfig",
59
+ "batch_size": 1,
60
+ "capacity_factor": null,
61
+ "checkpoint_id": "Qwen/Qwen3-Embedding-4B",
62
+ "checkpoint_revision": "5cf2132abc99cad020ac570b19d031efec650f2b",
63
+ "continuous_batching": false,
64
+ "ep_degree": 1,
65
+ "fused_qkv": true,
66
+ "glu_mlp": true,
67
+ "local_ranks_size": 2,
68
+ "max_batch_size": 1,
69
+ "max_context_length": 4096,
70
+ "max_topk": 256,
71
+ "n_active_tokens": 4096,
72
+ "neuronxcc_version": "2.21.33363.0+82129205",
73
+ "on_device_sampling": false,
74
+ "optimum_neuron_version": "0.4.6.dev2",
75
+ "output_logits": false,
76
+ "pp_degree": 1,
77
+ "sequence_length": 4096,
78
+ "speculation_length": 0,
79
+ "start_rank_id": 0,
80
+ "target": "trn1",
81
+ "torch_dtype": "bfloat16",
82
+ "tp_degree": 2
83
+ },
84
+ "num_attention_heads": 32,
85
+ "num_hidden_layers": 36,
86
+ "num_key_value_heads": 8,
87
+ "rms_norm_eps": 1e-06,
88
+ "rope_scaling": null,
89
+ "rope_theta": 1000000,
90
+ "sliding_window": null,
91
+ "tie_word_embeddings": true,
92
+ "use_cache": true,
93
+ "use_sliding_window": false,
94
+ "vocab_size": 151665
95
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.6.dev2/qwen3/Qwen/Qwen3-Embedding-4B/80514be9a02fddc9c476.json ADDED
@@ -0,0 +1,95 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "Qwen/Qwen3-Embedding-4B",
4
+ "_task": "feature-extraction",
5
+ "architectures": [
6
+ "Qwen3ForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "dtype": "bfloat16",
11
+ "head_dim": 128,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 2560,
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 9728,
16
+ "layer_types": [
17
+ "full_attention",
18
+ "full_attention",
19
+ "full_attention",
20
+ "full_attention",
21
+ "full_attention",
22
+ "full_attention",
23
+ "full_attention",
24
+ "full_attention",
25
+ "full_attention",
26
+ "full_attention",
27
+ "full_attention",
28
+ "full_attention",
29
+ "full_attention",
30
+ "full_attention",
31
+ "full_attention",
32
+ "full_attention",
33
+ "full_attention",
34
+ "full_attention",
35
+ "full_attention",
36
+ "full_attention",
37
+ "full_attention",
38
+ "full_attention",
39
+ "full_attention",
40
+ "full_attention",
41
+ "full_attention",
42
+ "full_attention",
43
+ "full_attention",
44
+ "full_attention",
45
+ "full_attention",
46
+ "full_attention",
47
+ "full_attention",
48
+ "full_attention",
49
+ "full_attention",
50
+ "full_attention",
51
+ "full_attention",
52
+ "full_attention"
53
+ ],
54
+ "max_position_embeddings": 40960,
55
+ "max_window_layers": 36,
56
+ "model_type": "qwen3",
57
+ "neuron": {
58
+ "_serialized_key": "NxDNeuronConfig",
59
+ "batch_size": 1,
60
+ "capacity_factor": null,
61
+ "checkpoint_id": "Qwen/Qwen3-Embedding-4B",
62
+ "checkpoint_revision": "5cf2132abc99cad020ac570b19d031efec650f2b",
63
+ "continuous_batching": false,
64
+ "ep_degree": 1,
65
+ "fused_qkv": true,
66
+ "glu_mlp": true,
67
+ "local_ranks_size": 2,
68
+ "max_batch_size": 1,
69
+ "max_context_length": 4096,
70
+ "max_topk": 256,
71
+ "n_active_tokens": 4096,
72
+ "neuronxcc_version": "2.21.33363.0+82129205",
73
+ "on_device_sampling": false,
74
+ "optimum_neuron_version": "0.4.6.dev2",
75
+ "output_logits": false,
76
+ "pp_degree": 1,
77
+ "sequence_length": 4096,
78
+ "speculation_length": 0,
79
+ "start_rank_id": 0,
80
+ "target": "trn1",
81
+ "torch_dtype": "bfloat16",
82
+ "tp_degree": 2
83
+ },
84
+ "num_attention_heads": 32,
85
+ "num_hidden_layers": 36,
86
+ "num_key_value_heads": 8,
87
+ "rms_norm_eps": 1e-06,
88
+ "rope_scaling": null,
89
+ "rope_theta": 1000000,
90
+ "sliding_window": null,
91
+ "tie_word_embeddings": true,
92
+ "use_cache": true,
93
+ "use_sliding_window": false,
94
+ "vocab_size": 151665
95
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.6.dev3/929b02754a13cbfdf657d863c3fc6f3bce672879bc6ae48ab45be21e881e9ec2/96baef0f2a01e5e29193.json ADDED
@@ -0,0 +1,88 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "Qwen/Qwen3-0.6B",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "Qwen3ForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "dtype": "bfloat16",
11
+ "head_dim": 128,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 1024,
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 3072,
16
+ "layer_types": [
17
+ "full_attention",
18
+ "full_attention",
19
+ "full_attention",
20
+ "full_attention",
21
+ "full_attention",
22
+ "full_attention",
23
+ "full_attention",
24
+ "full_attention",
25
+ "full_attention",
26
+ "full_attention",
27
+ "full_attention",
28
+ "full_attention",
29
+ "full_attention",
30
+ "full_attention",
31
+ "full_attention",
32
+ "full_attention",
33
+ "full_attention",
34
+ "full_attention",
35
+ "full_attention",
36
+ "full_attention",
37
+ "full_attention",
38
+ "full_attention",
39
+ "full_attention",
40
+ "full_attention",
41
+ "full_attention",
42
+ "full_attention",
43
+ "full_attention",
44
+ "full_attention"
45
+ ],
46
+ "max_position_embeddings": 40960,
47
+ "max_window_layers": 28,
48
+ "model_type": "qwen3",
49
+ "neuron": {
50
+ "_serialized_key": "NxDNeuronConfig",
51
+ "batch_size": 4,
52
+ "capacity_factor": null,
53
+ "checkpoint_id": "Qwen/Qwen3-0.6B",
54
+ "checkpoint_revision": "c1899de289a04d12100db370d81485cdf75e47ca",
55
+ "continuous_batching": true,
56
+ "ep_degree": 1,
57
+ "fused_qkv": true,
58
+ "glu_mlp": true,
59
+ "local_ranks_size": 1,
60
+ "max_batch_size": 4,
61
+ "max_context_length": 1024,
62
+ "max_topk": 256,
63
+ "n_active_tokens": 1024,
64
+ "neuronxcc_version": "2.21.33363.0+82129205",
65
+ "on_device_sampling": true,
66
+ "optimum_neuron_version": "0.4.6.dev3",
67
+ "output_logits": false,
68
+ "pp_degree": 1,
69
+ "prefill_chunk_size": 0,
70
+ "sequence_length": 1024,
71
+ "speculation_length": 0,
72
+ "start_rank_id": 0,
73
+ "target": "trn1",
74
+ "torch_dtype": "bfloat16",
75
+ "tp_degree": 1
76
+ },
77
+ "num_attention_heads": 16,
78
+ "num_hidden_layers": 28,
79
+ "num_key_value_heads": 8,
80
+ "rms_norm_eps": 1e-06,
81
+ "rope_scaling": null,
82
+ "rope_theta": 1000000,
83
+ "sliding_window": null,
84
+ "tie_word_embeddings": true,
85
+ "use_cache": true,
86
+ "use_sliding_window": false,
87
+ "vocab_size": 151936
88
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.6.dev3/c54cbc73d9a74e547bf7ca1feb2b290b641ed261e32f7c08baba5633884f1298/d73c51dc7dd75010abe9.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "meta-llama/Meta-Llama-3.1-8B-Instruct",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "LlamaForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "dtype": "bfloat16",
11
+ "head_dim": 128,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 4096,
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 14336,
16
+ "max_position_embeddings": 131072,
17
+ "mlp_bias": false,
18
+ "model_type": "llama",
19
+ "neuron": {
20
+ "_serialized_key": "NxDNeuronConfig",
21
+ "batch_size": 32,
22
+ "capacity_factor": null,
23
+ "checkpoint_id": "meta-llama/Meta-Llama-3.1-8B-Instruct",
24
+ "checkpoint_revision": "0e9e39f249a16976918f6564b8830bc894c89659",
25
+ "continuous_batching": true,
26
+ "ep_degree": 1,
27
+ "fused_qkv": true,
28
+ "glu_mlp": true,
29
+ "local_ranks_size": 8,
30
+ "max_batch_size": 32,
31
+ "max_context_length": 4096,
32
+ "max_topk": 256,
33
+ "n_active_tokens": 4096,
34
+ "neuronxcc_version": "2.21.33363.0+82129205",
35
+ "on_device_sampling": true,
36
+ "optimum_neuron_version": "0.4.6.dev3",
37
+ "output_logits": false,
38
+ "pp_degree": 1,
39
+ "prefill_chunk_size": 1024,
40
+ "sequence_length": 4096,
41
+ "speculation_length": 0,
42
+ "start_rank_id": 0,
43
+ "target": "trn1",
44
+ "torch_dtype": "bfloat16",
45
+ "tp_degree": 8
46
+ },
47
+ "num_attention_heads": 32,
48
+ "num_hidden_layers": 32,
49
+ "num_key_value_heads": 8,
50
+ "pretraining_tp": 1,
51
+ "rms_norm_eps": 1e-05,
52
+ "rope_scaling": {
53
+ "factor": 8.0,
54
+ "high_freq_factor": 4.0,
55
+ "low_freq_factor": 1.0,
56
+ "original_max_position_embeddings": 8192,
57
+ "rope_type": "llama3"
58
+ },
59
+ "rope_theta": 500000.0,
60
+ "tie_word_embeddings": false,
61
+ "use_cache": true,
62
+ "vocab_size": 128256
63
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.6.dev3/cf6b9a360dcf294104671106bae2adbd9fd291823bb60a351883163684073231/d182199b5d53c98ec562.json ADDED
@@ -0,0 +1,64 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "unsloth/Llama-3.2-1B-Instruct",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "LlamaForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "dtype": "bfloat16",
11
+ "head_dim": 64,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 2048,
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 8192,
16
+ "max_position_embeddings": 131072,
17
+ "mlp_bias": false,
18
+ "model_type": "llama",
19
+ "neuron": {
20
+ "_serialized_key": "NxDNeuronConfig",
21
+ "batch_size": 2,
22
+ "capacity_factor": null,
23
+ "checkpoint_id": "unsloth/Llama-3.2-1B-Instruct",
24
+ "checkpoint_revision": "5a8abab4a5d6f164389b1079fb721cfab8d7126c",
25
+ "continuous_batching": true,
26
+ "ep_degree": 1,
27
+ "fused_qkv": true,
28
+ "glu_mlp": true,
29
+ "local_ranks_size": 2,
30
+ "max_batch_size": 2,
31
+ "max_context_length": 512,
32
+ "max_topk": 256,
33
+ "n_active_tokens": 512,
34
+ "neuronxcc_version": "2.21.33363.0+82129205",
35
+ "on_device_sampling": true,
36
+ "optimum_neuron_version": "0.4.6.dev3",
37
+ "output_logits": false,
38
+ "pp_degree": 1,
39
+ "prefill_chunk_size": 0,
40
+ "sequence_length": 512,
41
+ "speculation_length": 0,
42
+ "start_rank_id": 0,
43
+ "target": "trn1",
44
+ "torch_dtype": "bfloat16",
45
+ "tp_degree": 2
46
+ },
47
+ "num_attention_heads": 32,
48
+ "num_hidden_layers": 16,
49
+ "num_key_value_heads": 8,
50
+ "pretraining_tp": 1,
51
+ "rms_norm_eps": 1e-05,
52
+ "rope_scaling": {
53
+ "factor": 32.0,
54
+ "high_freq_factor": 4.0,
55
+ "low_freq_factor": 1.0,
56
+ "original_max_position_embeddings": 8192,
57
+ "rope_type": "llama3"
58
+ },
59
+ "rope_theta": 500000.0,
60
+ "tie_word_embeddings": true,
61
+ "unsloth_fixed": true,
62
+ "use_cache": true,
63
+ "vocab_size": 128256
64
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.6.dev3/e0b6d1e2424243dcd9ff1755e02969dcc312d14df531d876c5c2892f285b2863/24e7f0a205508b46b0eb.json ADDED
@@ -0,0 +1,96 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "Qwen/Qwen3-Embedding-4B",
4
+ "_task": "feature-extraction",
5
+ "architectures": [
6
+ "Qwen3ForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "dtype": "bfloat16",
11
+ "head_dim": 128,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 2560,
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 9728,
16
+ "layer_types": [
17
+ "full_attention",
18
+ "full_attention",
19
+ "full_attention",
20
+ "full_attention",
21
+ "full_attention",
22
+ "full_attention",
23
+ "full_attention",
24
+ "full_attention",
25
+ "full_attention",
26
+ "full_attention",
27
+ "full_attention",
28
+ "full_attention",
29
+ "full_attention",
30
+ "full_attention",
31
+ "full_attention",
32
+ "full_attention",
33
+ "full_attention",
34
+ "full_attention",
35
+ "full_attention",
36
+ "full_attention",
37
+ "full_attention",
38
+ "full_attention",
39
+ "full_attention",
40
+ "full_attention",
41
+ "full_attention",
42
+ "full_attention",
43
+ "full_attention",
44
+ "full_attention",
45
+ "full_attention",
46
+ "full_attention",
47
+ "full_attention",
48
+ "full_attention",
49
+ "full_attention",
50
+ "full_attention",
51
+ "full_attention",
52
+ "full_attention"
53
+ ],
54
+ "max_position_embeddings": 40960,
55
+ "max_window_layers": 36,
56
+ "model_type": "qwen3",
57
+ "neuron": {
58
+ "_serialized_key": "NxDNeuronConfig",
59
+ "batch_size": 1,
60
+ "capacity_factor": null,
61
+ "checkpoint_id": "Qwen/Qwen3-Embedding-4B",
62
+ "checkpoint_revision": "5cf2132abc99cad020ac570b19d031efec650f2b",
63
+ "continuous_batching": false,
64
+ "ep_degree": 1,
65
+ "fused_qkv": true,
66
+ "glu_mlp": true,
67
+ "local_ranks_size": 2,
68
+ "max_batch_size": 1,
69
+ "max_context_length": 1024,
70
+ "max_topk": 256,
71
+ "n_active_tokens": 1024,
72
+ "neuronxcc_version": "2.21.33363.0+82129205",
73
+ "on_device_sampling": false,
74
+ "optimum_neuron_version": "0.4.6.dev3",
75
+ "output_logits": false,
76
+ "pp_degree": 1,
77
+ "prefill_chunk_size": 0,
78
+ "sequence_length": 1024,
79
+ "speculation_length": 0,
80
+ "start_rank_id": 0,
81
+ "target": "trn1",
82
+ "torch_dtype": "bfloat16",
83
+ "tp_degree": 2
84
+ },
85
+ "num_attention_heads": 32,
86
+ "num_hidden_layers": 36,
87
+ "num_key_value_heads": 8,
88
+ "rms_norm_eps": 1e-06,
89
+ "rope_scaling": null,
90
+ "rope_theta": 1000000,
91
+ "sliding_window": null,
92
+ "tie_word_embeddings": true,
93
+ "use_cache": true,
94
+ "use_sliding_window": false,
95
+ "vocab_size": 151665
96
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.6.dev3/e0b6d1e2424243dcd9ff1755e02969dcc312d14df531d876c5c2892f285b2863/6e2489df1bc1e5c5af63.json ADDED
@@ -0,0 +1,96 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "Qwen/Qwen3-Embedding-4B",
4
+ "_task": "feature-extraction",
5
+ "architectures": [
6
+ "Qwen3ForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "dtype": "bfloat16",
11
+ "head_dim": 128,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 2560,
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 9728,
16
+ "layer_types": [
17
+ "full_attention",
18
+ "full_attention",
19
+ "full_attention",
20
+ "full_attention",
21
+ "full_attention",
22
+ "full_attention",
23
+ "full_attention",
24
+ "full_attention",
25
+ "full_attention",
26
+ "full_attention",
27
+ "full_attention",
28
+ "full_attention",
29
+ "full_attention",
30
+ "full_attention",
31
+ "full_attention",
32
+ "full_attention",
33
+ "full_attention",
34
+ "full_attention",
35
+ "full_attention",
36
+ "full_attention",
37
+ "full_attention",
38
+ "full_attention",
39
+ "full_attention",
40
+ "full_attention",
41
+ "full_attention",
42
+ "full_attention",
43
+ "full_attention",
44
+ "full_attention",
45
+ "full_attention",
46
+ "full_attention",
47
+ "full_attention",
48
+ "full_attention",
49
+ "full_attention",
50
+ "full_attention",
51
+ "full_attention",
52
+ "full_attention"
53
+ ],
54
+ "max_position_embeddings": 40960,
55
+ "max_window_layers": 36,
56
+ "model_type": "qwen3",
57
+ "neuron": {
58
+ "_serialized_key": "NxDNeuronConfig",
59
+ "batch_size": 1,
60
+ "capacity_factor": null,
61
+ "checkpoint_id": "Qwen/Qwen3-Embedding-4B",
62
+ "checkpoint_revision": "5cf2132abc99cad020ac570b19d031efec650f2b",
63
+ "continuous_batching": false,
64
+ "ep_degree": 1,
65
+ "fused_qkv": true,
66
+ "glu_mlp": true,
67
+ "local_ranks_size": 2,
68
+ "max_batch_size": 1,
69
+ "max_context_length": 4096,
70
+ "max_topk": 256,
71
+ "n_active_tokens": 4096,
72
+ "neuronxcc_version": "2.21.33363.0+82129205",
73
+ "on_device_sampling": false,
74
+ "optimum_neuron_version": "0.4.6.dev3",
75
+ "output_logits": false,
76
+ "pp_degree": 1,
77
+ "prefill_chunk_size": 1024,
78
+ "sequence_length": 4096,
79
+ "speculation_length": 0,
80
+ "start_rank_id": 0,
81
+ "target": "trn1",
82
+ "torch_dtype": "bfloat16",
83
+ "tp_degree": 2
84
+ },
85
+ "num_attention_heads": 32,
86
+ "num_hidden_layers": 36,
87
+ "num_key_value_heads": 8,
88
+ "rms_norm_eps": 1e-06,
89
+ "rope_scaling": null,
90
+ "rope_theta": 1000000,
91
+ "sliding_window": null,
92
+ "tie_word_embeddings": true,
93
+ "use_cache": true,
94
+ "use_sliding_window": false,
95
+ "vocab_size": 151665
96
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.6.dev3/e0b6d1e2424243dcd9ff1755e02969dcc312d14df531d876c5c2892f285b2863/a34b748f9038b1e376dc.json ADDED
@@ -0,0 +1,96 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "Qwen/Qwen3-Embedding-4B",
4
+ "_task": "feature-extraction",
5
+ "architectures": [
6
+ "Qwen3ForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "dtype": "bfloat16",
11
+ "head_dim": 128,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 2560,
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 9728,
16
+ "layer_types": [
17
+ "full_attention",
18
+ "full_attention",
19
+ "full_attention",
20
+ "full_attention",
21
+ "full_attention",
22
+ "full_attention",
23
+ "full_attention",
24
+ "full_attention",
25
+ "full_attention",
26
+ "full_attention",
27
+ "full_attention",
28
+ "full_attention",
29
+ "full_attention",
30
+ "full_attention",
31
+ "full_attention",
32
+ "full_attention",
33
+ "full_attention",
34
+ "full_attention",
35
+ "full_attention",
36
+ "full_attention",
37
+ "full_attention",
38
+ "full_attention",
39
+ "full_attention",
40
+ "full_attention",
41
+ "full_attention",
42
+ "full_attention",
43
+ "full_attention",
44
+ "full_attention",
45
+ "full_attention",
46
+ "full_attention",
47
+ "full_attention",
48
+ "full_attention",
49
+ "full_attention",
50
+ "full_attention",
51
+ "full_attention",
52
+ "full_attention"
53
+ ],
54
+ "max_position_embeddings": 40960,
55
+ "max_window_layers": 36,
56
+ "model_type": "qwen3",
57
+ "neuron": {
58
+ "_serialized_key": "NxDNeuronConfig",
59
+ "batch_size": 1,
60
+ "capacity_factor": null,
61
+ "checkpoint_id": "Qwen/Qwen3-Embedding-4B",
62
+ "checkpoint_revision": "5cf2132abc99cad020ac570b19d031efec650f2b",
63
+ "continuous_batching": false,
64
+ "ep_degree": 1,
65
+ "fused_qkv": true,
66
+ "glu_mlp": true,
67
+ "local_ranks_size": 2,
68
+ "max_batch_size": 1,
69
+ "max_context_length": 2048,
70
+ "max_topk": 256,
71
+ "n_active_tokens": 2048,
72
+ "neuronxcc_version": "2.21.33363.0+82129205",
73
+ "on_device_sampling": false,
74
+ "optimum_neuron_version": "0.4.6.dev3",
75
+ "output_logits": false,
76
+ "pp_degree": 1,
77
+ "prefill_chunk_size": 1024,
78
+ "sequence_length": 2048,
79
+ "speculation_length": 0,
80
+ "start_rank_id": 0,
81
+ "target": "trn1",
82
+ "torch_dtype": "bfloat16",
83
+ "tp_degree": 2
84
+ },
85
+ "num_attention_heads": 32,
86
+ "num_hidden_layers": 36,
87
+ "num_key_value_heads": 8,
88
+ "rms_norm_eps": 1e-06,
89
+ "rope_scaling": null,
90
+ "rope_theta": 1000000,
91
+ "sliding_window": null,
92
+ "tie_word_embeddings": true,
93
+ "use_cache": true,
94
+ "use_sliding_window": false,
95
+ "vocab_size": 151665
96
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.6.dev3/llama/meta-llama/Meta-Llama-3.1-8B-Instruct/d73c51dc7dd75010abe9.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "meta-llama/Meta-Llama-3.1-8B-Instruct",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "LlamaForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "dtype": "bfloat16",
11
+ "head_dim": 128,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 4096,
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 14336,
16
+ "max_position_embeddings": 131072,
17
+ "mlp_bias": false,
18
+ "model_type": "llama",
19
+ "neuron": {
20
+ "_serialized_key": "NxDNeuronConfig",
21
+ "batch_size": 32,
22
+ "capacity_factor": null,
23
+ "checkpoint_id": "meta-llama/Meta-Llama-3.1-8B-Instruct",
24
+ "checkpoint_revision": "0e9e39f249a16976918f6564b8830bc894c89659",
25
+ "continuous_batching": true,
26
+ "ep_degree": 1,
27
+ "fused_qkv": true,
28
+ "glu_mlp": true,
29
+ "local_ranks_size": 8,
30
+ "max_batch_size": 32,
31
+ "max_context_length": 4096,
32
+ "max_topk": 256,
33
+ "n_active_tokens": 4096,
34
+ "neuronxcc_version": "2.21.33363.0+82129205",
35
+ "on_device_sampling": true,
36
+ "optimum_neuron_version": "0.4.6.dev3",
37
+ "output_logits": false,
38
+ "pp_degree": 1,
39
+ "prefill_chunk_size": 1024,
40
+ "sequence_length": 4096,
41
+ "speculation_length": 0,
42
+ "start_rank_id": 0,
43
+ "target": "trn1",
44
+ "torch_dtype": "bfloat16",
45
+ "tp_degree": 8
46
+ },
47
+ "num_attention_heads": 32,
48
+ "num_hidden_layers": 32,
49
+ "num_key_value_heads": 8,
50
+ "pretraining_tp": 1,
51
+ "rms_norm_eps": 1e-05,
52
+ "rope_scaling": {
53
+ "factor": 8.0,
54
+ "high_freq_factor": 4.0,
55
+ "low_freq_factor": 1.0,
56
+ "original_max_position_embeddings": 8192,
57
+ "rope_type": "llama3"
58
+ },
59
+ "rope_theta": 500000.0,
60
+ "tie_word_embeddings": false,
61
+ "use_cache": true,
62
+ "vocab_size": 128256
63
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.6.dev3/qwen3/Qwen/Qwen3-Embedding-4B/24e7f0a205508b46b0eb.json ADDED
@@ -0,0 +1,96 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "Qwen/Qwen3-Embedding-4B",
4
+ "_task": "feature-extraction",
5
+ "architectures": [
6
+ "Qwen3ForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "dtype": "bfloat16",
11
+ "head_dim": 128,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 2560,
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 9728,
16
+ "layer_types": [
17
+ "full_attention",
18
+ "full_attention",
19
+ "full_attention",
20
+ "full_attention",
21
+ "full_attention",
22
+ "full_attention",
23
+ "full_attention",
24
+ "full_attention",
25
+ "full_attention",
26
+ "full_attention",
27
+ "full_attention",
28
+ "full_attention",
29
+ "full_attention",
30
+ "full_attention",
31
+ "full_attention",
32
+ "full_attention",
33
+ "full_attention",
34
+ "full_attention",
35
+ "full_attention",
36
+ "full_attention",
37
+ "full_attention",
38
+ "full_attention",
39
+ "full_attention",
40
+ "full_attention",
41
+ "full_attention",
42
+ "full_attention",
43
+ "full_attention",
44
+ "full_attention",
45
+ "full_attention",
46
+ "full_attention",
47
+ "full_attention",
48
+ "full_attention",
49
+ "full_attention",
50
+ "full_attention",
51
+ "full_attention",
52
+ "full_attention"
53
+ ],
54
+ "max_position_embeddings": 40960,
55
+ "max_window_layers": 36,
56
+ "model_type": "qwen3",
57
+ "neuron": {
58
+ "_serialized_key": "NxDNeuronConfig",
59
+ "batch_size": 1,
60
+ "capacity_factor": null,
61
+ "checkpoint_id": "Qwen/Qwen3-Embedding-4B",
62
+ "checkpoint_revision": "5cf2132abc99cad020ac570b19d031efec650f2b",
63
+ "continuous_batching": false,
64
+ "ep_degree": 1,
65
+ "fused_qkv": true,
66
+ "glu_mlp": true,
67
+ "local_ranks_size": 2,
68
+ "max_batch_size": 1,
69
+ "max_context_length": 1024,
70
+ "max_topk": 256,
71
+ "n_active_tokens": 1024,
72
+ "neuronxcc_version": "2.21.33363.0+82129205",
73
+ "on_device_sampling": false,
74
+ "optimum_neuron_version": "0.4.6.dev3",
75
+ "output_logits": false,
76
+ "pp_degree": 1,
77
+ "prefill_chunk_size": 0,
78
+ "sequence_length": 1024,
79
+ "speculation_length": 0,
80
+ "start_rank_id": 0,
81
+ "target": "trn1",
82
+ "torch_dtype": "bfloat16",
83
+ "tp_degree": 2
84
+ },
85
+ "num_attention_heads": 32,
86
+ "num_hidden_layers": 36,
87
+ "num_key_value_heads": 8,
88
+ "rms_norm_eps": 1e-06,
89
+ "rope_scaling": null,
90
+ "rope_theta": 1000000,
91
+ "sliding_window": null,
92
+ "tie_word_embeddings": true,
93
+ "use_cache": true,
94
+ "use_sliding_window": false,
95
+ "vocab_size": 151665
96
+ }
neuronxcc-2.21.33363.0+82129205/MODULE_0352823045bfbd34fd3d+fb4cc044/compile_flags.json ADDED
@@ -0,0 +1 @@
 
 
1
+ ["--target=trn1", "--auto-cast=none", "--model-type=transformer", "--tensorizer-options=--enable-ccop-compute-overlap --cc-pipeline-tiling-factor=2 --vectorize-strided-dma ", "-O2", "--lnc=1", "--logfile=/tmp/nxd_model/encoding/_tp0_bk0/log-neuron-cc.txt"]
neuronxcc-2.21.33363.0+82129205/MODULE_0352823045bfbd34fd3d+fb4cc044/model.done ADDED
File without changes
neuronxcc-2.21.33363.0+82129205/MODULE_0352823045bfbd34fd3d+fb4cc044/model.hlo_module.pb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1418a3bc01d777bbefb8b367e96770bb02210c82476f56c4d78293cc1a7ea8bc
3
+ size 848773
neuronxcc-2.21.33363.0+82129205/MODULE_0352823045bfbd34fd3d+fb4cc044/model.neff ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6bc69ce1b547d85de7a704e3830e155358ac4147366aab81cd690f7274405844
3
+ size 4619264
neuronxcc-2.21.33363.0+82129205/MODULE_04dc6de9513dd8bb05c4+fb4cc044/compile_flags.json ADDED
@@ -0,0 +1 @@
 
 
1
+ ["--target=trn1", "--auto-cast=none", "--model-type=transformer", "--tensorizer-options=--enable-ccop-compute-overlap --cc-pipeline-tiling-factor=2 --vectorize-strided-dma ", "-O2", "--lnc=1", "--logfile=/tmp/nxd_model/encoding/_tp0_bk0/log-neuron-cc.txt"]
neuronxcc-2.21.33363.0+82129205/MODULE_04dc6de9513dd8bb05c4+fb4cc044/model.done ADDED
File without changes
neuronxcc-2.21.33363.0+82129205/MODULE_04dc6de9513dd8bb05c4+fb4cc044/model.hlo_module.pb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c92837716166f7cb39c6297bc51bacaccdd8471f7a7cf4eb5a90aa139ca3f28b
3
+ size 628810
neuronxcc-2.21.33363.0+82129205/MODULE_04dc6de9513dd8bb05c4+fb4cc044/model.neff ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e794bd527bf159e760f7e574be86936f0708fb508abd78488a3d208f0ae63803
3
+ size 22672384
neuronxcc-2.21.33363.0+82129205/MODULE_1759fced9dd9a678973e+24129607/compile_flags.json ADDED
@@ -0,0 +1 @@
 
 
1
+ ["--target=trn1", "--auto-cast=none", "--model-type=transformer", "--tensorizer-options=--enable-ccop-compute-overlap --cc-pipeline-tiling-factor=2 --vectorize-strided-dma ", "-O2", "--lnc=1", "--logfile=/tmp/nxd_model/context_encoding/_tp0_bk0/log-neuron-cc.txt"]
neuronxcc-2.21.33363.0+82129205/MODULE_1759fced9dd9a678973e+24129607/model.done ADDED
File without changes
neuronxcc-2.21.33363.0+82129205/MODULE_1759fced9dd9a678973e+24129607/model.hlo_module.pb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8223a0880f04437df80f89de1868b68be53988eb9a109c31fbf2423e0595ab55
3
+ size 684041
neuronxcc-2.21.33363.0+82129205/MODULE_1759fced9dd9a678973e+24129607/model.neff ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:31b87a1ef2e12d7cf6675e391738773cafeccfb746d2312644814bef9e634af2
3
+ size 7732224
neuronxcc-2.21.33363.0+82129205/MODULE_1824b665798bf04e2aa2+24129607/compile_flags.json ADDED
@@ -0,0 +1 @@
 
 
1
+ ["--target=trn1", "--auto-cast=none", "--model-type=transformer", "--tensorizer-options=--enable-ccop-compute-overlap --cc-pipeline-tiling-factor=2 --vectorize-strided-dma ", "-O2", "--lnc=1", "--logfile=/tmp/nxd_model/context_encoding/_tp0_bk0/log-neuron-cc.txt"]
neuronxcc-2.21.33363.0+82129205/MODULE_1824b665798bf04e2aa2+24129607/model.done ADDED
File without changes
neuronxcc-2.21.33363.0+82129205/MODULE_1824b665798bf04e2aa2+24129607/model.hlo_module.pb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b1eab0e1e36ccc80e01b9fc5973caceec04c50f7f81d61ab27925c6145addeec
3
+ size 412156
neuronxcc-2.21.33363.0+82129205/MODULE_1824b665798bf04e2aa2+24129607/model.neff ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8cde39dbd1fb71e45804b6ac56f97815205e6a7d6a7deac0af9414593434681b
3
+ size 2796544
neuronxcc-2.21.33363.0+82129205/MODULE_24a7a312e7cf7a6b6c5d+fb4cc044/compile_flags.json ADDED
@@ -0,0 +1 @@
 
 
1
+ ["--target=trn1", "--auto-cast=none", "--model-type=transformer", "--tensorizer-options=--enable-ccop-compute-overlap --cc-pipeline-tiling-factor=2 --vectorize-strided-dma ", "-O2", "--lnc=1", "--logfile=/tmp/nxd_model/encoding/_tp0_bk0/log-neuron-cc.txt"]
neuronxcc-2.21.33363.0+82129205/MODULE_24a7a312e7cf7a6b6c5d+fb4cc044/model.done ADDED
File without changes
neuronxcc-2.21.33363.0+82129205/MODULE_24a7a312e7cf7a6b6c5d+fb4cc044/model.hlo_module.pb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4d79d2330194768dee128be01cd0cbea2a63911f50e0c35b64ea4960ea48c493
3
+ size 848917
neuronxcc-2.21.33363.0+82129205/MODULE_24a7a312e7cf7a6b6c5d+fb4cc044/model.neff ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4e1571335377de3b6f476c55def0ba070f37fddb4a06253cf0f8b14da0e22e80
3
+ size 11254784
neuronxcc-2.21.33363.0+82129205/MODULE_68b1b688bc907867445e+a02c3a36/model.hlo_module.pb CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:1b82959150728b621645262435120008706a4ef119a943f7db4a6d90aeb430c7
3
  size 728309
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9d89321cf197c8909f26412855cae21850d77301d0ab013b344fb30575a4a7b5
3
  size 728309
neuronxcc-2.21.33363.0+82129205/MODULE_68b1b688bc907867445e+a02c3a36/model.neff CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:6ed190bc44698b591685d8b6feb0da87c5d1629abe304ab17569a3510a1a0fab
3
  size 7117824
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d7dff02b9d351d1d8c9461f01f9409fd21e2e118d101112e6e577cdc79664333
3
  size 7117824
neuronxcc-2.21.33363.0+82129205/MODULE_68b1b688bc907867445e+a02c3a36/wrapped_neff.hlo CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:0e34e480fdf7a4270bca4077b8ded8f08218d466c960d9cd822ee3690dc8e334
3
  size 7264840
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f90ab0964874ee64d7464bea3df754dd92e3a9b03a300d19dc9db7063702dbb1
3
  size 7264840
neuronxcc-2.21.33363.0+82129205/MODULE_88833531cb6c582d4830+a02c3a36/compile_flags.json ADDED
@@ -0,0 +1 @@
 
 
1
+ ["--target=trn1", "--auto-cast=none", "--model-type=transformer", "--tensorizer-options=--enable-ccop-compute-overlap --cc-pipeline-tiling-factor=2 --vectorize-strided-dma ", "-O2", "--lnc=1", "--logfile=/tmp/nxd_model/token_generation/_tp0_bk0/log-neuron-cc.txt", "--enable-internal-neff-wrapper"]
neuronxcc-2.21.33363.0+82129205/MODULE_88833531cb6c582d4830+a02c3a36/model.done ADDED
File without changes
neuronxcc-2.21.33363.0+82129205/MODULE_88833531cb6c582d4830+a02c3a36/model.hlo_module.pb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a344a8bfb0bd70d150a25d375f0a7ad1432403ec9c9a9455db46debc4a8e84eb
3
+ size 727217
neuronxcc-2.21.33363.0+82129205/MODULE_88833531cb6c582d4830+a02c3a36/model.neff ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:20e6a3e924fa9631c44d91b12bc5ea4e843680266ab0e765a4989e699c430fea
3
+ size 3472384
neuronxcc-2.21.33363.0+82129205/MODULE_88833531cb6c582d4830+a02c3a36/wrapped_neff.hlo ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bdacf6d0dbac59c6eb12d348bb6273380fbbc85b63605b48e5bff0b6d951e062
3
+ size 3610166
neuronxcc-2.21.33363.0+82129205/MODULE_9068b179aef4ffa33207+6c043b9a/compile_flags.json ADDED
@@ -0,0 +1 @@
 
 
1
+ ["--target=trn1", "--auto-cast=none", "--model-type=transformer", "--tensorizer-options=--enable-ccop-compute-overlap --cc-pipeline-tiling-factor=2 --vectorize-strided-dma ", "-O2", "--lnc=1", "--logfile=/tmp/nxd_model/chunked_encoding/_tp0_bk0/log-neuron-cc.txt"]
neuronxcc-2.21.33363.0+82129205/MODULE_9068b179aef4ffa33207+6c043b9a/model.hlo_module.pb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9ffa25d516e37232d7dd03d3456190d81b046f1be6c8899eb4d314ef378edd56
3
+ size 1171934
neuronxcc-2.21.33363.0+82129205/MODULE_92b96f10a3841c3c7919+6c043b9a/compile_flags.json ADDED
@@ -0,0 +1 @@
 
 
1
+ ["--target=trn1", "--auto-cast=none", "--model-type=transformer", "--tensorizer-options=--enable-ccop-compute-overlap --cc-pipeline-tiling-factor=2 --vectorize-strided-dma ", "-O2", "--lnc=1", "--logfile=/tmp/nxd_model/chunked_encoding/_tp0_bk0/log-neuron-cc.txt"]
neuronxcc-2.21.33363.0+82129205/MODULE_92b96f10a3841c3c7919+6c043b9a/model.done ADDED
File without changes
neuronxcc-2.21.33363.0+82129205/MODULE_92b96f10a3841c3c7919+6c043b9a/model.hlo_module.pb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b61cf92085320a14cfb797578aa3fb39f5cba66690c0019cba79b75a2b244a32
3
+ size 1165790
neuronxcc-2.21.33363.0+82129205/MODULE_92b96f10a3841c3c7919+6c043b9a/model.neff ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f5a5214416bbbf27c2666b320c779196f07ad6ebfd0ef705474d73bf5b4c3817
3
+ size 49859584
neuronxcc-2.21.33363.0+82129205/MODULE_95b9072f246645f24461+a02c3a36/model.hlo_module.pb CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:12bcc80b2ad2cb1e12abbca61cd3f324f6f4f564cf78d121d0f5fcd01305c683
3
  size 451319
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b364b15f88cdba27d3ac4973f7dcfb2cbc49da9abd971bc1f0c52315364d55dc
3
  size 451319
neuronxcc-2.21.33363.0+82129205/MODULE_95b9072f246645f24461+a02c3a36/model.neff CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:bdf04a70eb8bda2f7f4e49bedb2c346136c7e9b0dec32fa7421ea72474563fbe
3
  size 2509824
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f173e6c1f419b379767e720c82df3bec9c7942c49ca6ce29905fdea0ee240610
3
  size 2509824
neuronxcc-2.21.33363.0+82129205/MODULE_95b9072f246645f24461+a02c3a36/wrapped_neff.hlo CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:c84f3a91584fb71a18f3ba16dd9c1cde9917ff28ab72e432ab03a61ad722f8dc
3
  size 2583911
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5cde9a5e1a781062038054f87a06889f4898ef6a5b5a50ae52638bd62e04342a
3
  size 2583911
neuronxcc-2.21.33363.0+82129205/MODULE_a431a415019174ce0469+6c043b9a/compile_flags.json ADDED
@@ -0,0 +1 @@
 
 
1
+ ["--target=trn1", "--auto-cast=none", "--model-type=transformer", "--tensorizer-options=--enable-ccop-compute-overlap --cc-pipeline-tiling-factor=2 --vectorize-strided-dma ", "-O2", "--lnc=1", "--logfile=/tmp/nxd_model/chunked_encoding/_tp0_bk0/log-neuron-cc.txt"]