Hugging Face's logo

aws-neuron
/
optimum-neuron-cache

Model card Files
xet
Community
641
dacorvo HF Staff commited on
Commit
26297a6
·
verified ·
1 Parent(s): 29e118a

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 +12 -0
  2. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/0cc25526e2cfc37a8875a3752f33c4d7505d8a07b869d0f3f41915cf6e763b74/440dcb82274f114a66ad.json +58 -0
  3. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/0cc25526e2cfc37a8875a3752f33c4d7505d8a07b869d0f3f41915cf6e763b74/9bb0e82ff320f6e2c0bc.json +58 -0
  4. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/2da7a00f0478d50ae1e7f75f085c5b2773b5f355f427c61cf34cb6febd629d96/4925f3a03aeddc47881d.json +59 -0
  5. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/4ab8140bc7eb4a553d95855c5c2be2cf8c0fbab21b823d76183b6f51e98b6fc5/93ac0e7f7bd685a9d58b.json +58 -0
  6. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/4cb7aff9e2a15c151396f2b684013e39d6739f0dec83e5c9dabbfe9d5fcf77b7/34454e3efb576d636921.json +82 -0
  7. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/4cb7aff9e2a15c151396f2b684013e39d6739f0dec83e5c9dabbfe9d5fcf77b7/5ec0bcb03ab4e2ad844c.json +82 -0
  8. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/4cb7aff9e2a15c151396f2b684013e39d6739f0dec83e5c9dabbfe9d5fcf77b7/df54a6f9b56ca3bdfc41.json +82 -0
  9. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/4d281af1651143f9a83b42af25d9f0a533dbbd9c3441537127ae772fd3e399cc/2c2bad80f80ab55a2d84.json +87 -0
  10. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/4d281af1651143f9a83b42af25d9f0a533dbbd9c3441537127ae772fd3e399cc/9c8cb869b5d7e2540fa0.json +87 -0
  11. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/6454afdf3e9d66c7226c13a575b718845c25e53b0699600ba2bb4f883e9d841b/12ce6ef4377bb2672028.json +62 -0
  12. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/6454afdf3e9d66c7226c13a575b718845c25e53b0699600ba2bb4f883e9d841b/5997ab86579aadcc5cdd.json +62 -0
  13. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/6454afdf3e9d66c7226c13a575b718845c25e53b0699600ba2bb4f883e9d841b/f85a0d1bea066bc1274b.json +62 -0
  14. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/73707b485eab9008c7aba7f5dad0ce2384ac685318d5f888c12fa0d81ed90b19/1a86e1cfcaf229143d5c.json +134 -0
  15. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/73707b485eab9008c7aba7f5dad0ce2384ac685318d5f888c12fa0d81ed90b19/7144c43faca4fea18b01.json +134 -0
  16. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/73707b485eab9008c7aba7f5dad0ce2384ac685318d5f888c12fa0d81ed90b19/eb6ead215819eab7b04a.json +134 -0
  17. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/7518518c7e077820070186deda960d8cc49db068cdf0ac70664098fa2b6b698c/a037d3d5582033b493b7.json +64 -0
  18. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/8c90ac2593ed0b7f1ecb60e82cb184fb11f2ea640befa1cc7b10766a5c02525d/8cefa5ff9083418aa040.json +164 -0
  19. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/8c90ac2593ed0b7f1ecb60e82cb184fb11f2ea640befa1cc7b10766a5c02525d/9fbd66c92baabc8b5ad0.json +164 -0
  20. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/920f44ce6d3e004d1ce547ae06644f7be262180644b04573153aa15d98742edc/ada37737b07aadde24a9.json +65 -0
  21. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/929b02754a13cbfdf657d863c3fc6f3bce672879bc6ae48ab45be21e881e9ec2/97503f4e2dca15bd6721.json +87 -0
  22. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/9c24a2080ec26e9c74d5bd90866f74131aa4d5c975415b96f5e6600f081d5647/a7d7ccbe80fcc54f090d.json +81 -0
  23. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/cf6b9a360dcf294104671106bae2adbd9fd291823bb60a351883163684073231/0aa29066680da68d477e.json +63 -0
  24. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/cf6b9a360dcf294104671106bae2adbd9fd291823bb60a351883163684073231/4aab355cf64b87f1cff4.json +63 -0
  25. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/cf6b9a360dcf294104671106bae2adbd9fd291823bb60a351883163684073231/72eee6f943925a38948a.json +63 -0
  26. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/cf6b9a360dcf294104671106bae2adbd9fd291823bb60a351883163684073231/c4a5702b42ed466de9e7.json +63 -0
  27. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/cf6b9a360dcf294104671106bae2adbd9fd291823bb60a351883163684073231/f18d2067d0df4f48c0c4.json +63 -0
  28. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/d139acf64685f15794bb983ff6eb881bdd31304bae88b0ce1ed20a54c21f2265/2eeaf9c9a7977232515a.json +58 -0
  29. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/granite/hf-internal-testing/tiny-random-GraniteForCausalLM/2eeaf9c9a7977232515a.json +58 -0
  30. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/granite/ibm-granite/granite-3.1-2b-instruct/440dcb82274f114a66ad.json +58 -0
  31. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/llama/llamafactory/tiny-random-Llama-3/5997ab86579aadcc5cdd.json +62 -0
  32. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/llama/unsloth/Llama-3.2-1B-Instruct/4aab355cf64b87f1cff4.json +63 -0
  33. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/llama4_text/tiny-random/llama-4/a7d7ccbe80fcc54f090d.json +81 -0
  34. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/mixtral/dacorvo/Mixtral-tiny/93ac0e7f7bd685a9d58b.json +58 -0
  35. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/phi3/microsoft/Phi-3.5-mini-instruct/8cefa5ff9083418aa040.json +164 -0
  36. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/phi3/yujiepan/phi-4-tiny-random/4925f3a03aeddc47881d.json +59 -0
  37. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/qwen2/Qwen/Qwen2.5-0.5B/5ec0bcb03ab4e2ad844c.json +82 -0
  38. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/qwen2/yujiepan/qwen2.5-128k-tiny-random/a037d3d5582033b493b7.json +64 -0
  39. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/qwen3/Qwen/Qwen3-Embedding-0.6B/2c2bad80f80ab55a2d84.json +87 -0
  40. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/qwen3_moe/optimum-internal-testing/tiny-random-qwen3_moe/ada37737b07aadde24a9.json +65 -0
  41. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/smollm3/HuggingFaceTB/SmolLM3-3B/7144c43faca4fea18b01.json +134 -0
  42. neuronxcc-2.21.33363.0+82129205/MODULE_04d46d63afe5b4e1584b+6bc3ae29/compile_flags.json +1 -0
  43. neuronxcc-2.21.33363.0+82129205/MODULE_04d46d63afe5b4e1584b+6bc3ae29/model.done +0 -0
  44. neuronxcc-2.21.33363.0+82129205/MODULE_04d46d63afe5b4e1584b+6bc3ae29/model.hlo_module.pb +3 -0
  45. neuronxcc-2.21.33363.0+82129205/MODULE_04d46d63afe5b4e1584b+6bc3ae29/model.neff +3 -0
  46. neuronxcc-2.21.33363.0+82129205/MODULE_04d46d63afe5b4e1584b+6bc3ae29/wrapped_neff.hlo +3 -0
  47. neuronxcc-2.21.33363.0+82129205/MODULE_10217061096959125489+e30acd3a/model.neff +0 -0
  48. neuronxcc-2.21.33363.0+82129205/MODULE_10244305442015770634+e30acd3a/model.neff +0 -0
  49. neuronxcc-2.21.33363.0+82129205/MODULE_10606948783918825529+e30acd3a/model.neff +0 -0
  50. neuronxcc-2.21.33363.0+82129205/MODULE_10645643398657092095+e30acd3a/model.neff +0 -0
.gitattributes CHANGED
@@ -15552,3 +15552,15 @@ neuronxcc-2.21.33363.0+82129205/MODULE_f22fe1c9b82b6357a712+4a958caf/wrapped_nef
15552
  neuronxcc-2.21.33363.0+82129205/MODULE_35f89dd578b938fdd291+a02c3a36/model.neff filter=lfs diff=lfs merge=lfs -text
15553
  neuronxcc-2.21.33363.0+82129205/MODULE_35f89dd578b938fdd291+a02c3a36/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
15554
  neuronxcc-2.21.33363.0+82129205/MODULE_ab64b202fcda3e011f28+24129607/model.neff filter=lfs diff=lfs merge=lfs -text
 
 
 
 
 
 
 
 
 
 
 
 
 
15552
  neuronxcc-2.21.33363.0+82129205/MODULE_35f89dd578b938fdd291+a02c3a36/model.neff filter=lfs diff=lfs merge=lfs -text
15553
  neuronxcc-2.21.33363.0+82129205/MODULE_35f89dd578b938fdd291+a02c3a36/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
15554
  neuronxcc-2.21.33363.0+82129205/MODULE_ab64b202fcda3e011f28+24129607/model.neff filter=lfs diff=lfs merge=lfs -text
15555
+ neuronxcc-2.21.33363.0+82129205/MODULE_04d46d63afe5b4e1584b+6bc3ae29/model.neff filter=lfs diff=lfs merge=lfs -text
15556
+ neuronxcc-2.21.33363.0+82129205/MODULE_04d46d63afe5b4e1584b+6bc3ae29/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
15557
+ neuronxcc-2.21.33363.0+82129205/MODULE_311c80d1ee4083869ef9+35105abb/model.neff filter=lfs diff=lfs merge=lfs -text
15558
+ neuronxcc-2.21.33363.0+82129205/MODULE_311c80d1ee4083869ef9+35105abb/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
15559
+ neuronxcc-2.21.33363.0+82129205/MODULE_57eeb604bf270dafbec4+9e6e935c/model.neff filter=lfs diff=lfs merge=lfs -text
15560
+ neuronxcc-2.21.33363.0+82129205/MODULE_57eeb604bf270dafbec4+9e6e935c/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
15561
+ neuronxcc-2.21.33363.0+82129205/MODULE_812cbf042a260062f4b9+d1b57957/model.neff filter=lfs diff=lfs merge=lfs -text
15562
+ neuronxcc-2.21.33363.0+82129205/MODULE_812cbf042a260062f4b9+d1b57957/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
15563
+ neuronxcc-2.21.33363.0+82129205/MODULE_beeee808ee2c5817dadb+41bea4e3/model.neff filter=lfs diff=lfs merge=lfs -text
15564
+ neuronxcc-2.21.33363.0+82129205/MODULE_beeee808ee2c5817dadb+41bea4e3/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
15565
+ neuronxcc-2.21.33363.0+82129205/MODULE_c01f2cdb2ffaf3dd482e+0c4991c2/model.neff filter=lfs diff=lfs merge=lfs -text
15566
+ neuronxcc-2.21.33363.0+82129205/MODULE_c01f2cdb2ffaf3dd482e+0c4991c2/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/0cc25526e2cfc37a8875a3752f33c4d7505d8a07b869d0f3f41915cf6e763b74/440dcb82274f114a66ad.json ADDED
@@ -0,0 +1,58 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "ibm-granite/granite-3.1-2b-instruct",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "GraniteForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.1,
10
+ "attention_multiplier": 0.015625,
11
+ "dtype": "bfloat16",
12
+ "embedding_multiplier": 12.0,
13
+ "hidden_act": "silu",
14
+ "hidden_size": 2048,
15
+ "initializer_range": 0.02,
16
+ "intermediate_size": 8192,
17
+ "logits_scaling": 8.0,
18
+ "max_position_embeddings": 131072,
19
+ "mlp_bias": false,
20
+ "model_type": "granite",
21
+ "neuron": {
22
+ "_serialized_key": "NxDNeuronConfig",
23
+ "batch_size": 4,
24
+ "capacity_factor": null,
25
+ "checkpoint_id": "ibm-granite/granite-3.1-2b-instruct",
26
+ "checkpoint_revision": "bbc2aed595bd38bd770263dc3ab831db9794441d",
27
+ "continuous_batching": true,
28
+ "ep_degree": 1,
29
+ "fused_qkv": true,
30
+ "glu_mlp": true,
31
+ "local_ranks_size": 2,
32
+ "max_batch_size": 4,
33
+ "max_context_length": 4096,
34
+ "max_topk": 256,
35
+ "n_active_tokens": 4096,
36
+ "neuronxcc_version": "2.21.33363.0+82129205",
37
+ "on_device_sampling": true,
38
+ "optimum_neuron_version": "0.4.5.dev1",
39
+ "output_logits": false,
40
+ "pp_degree": 1,
41
+ "sequence_length": 4096,
42
+ "speculation_length": 0,
43
+ "start_rank_id": 0,
44
+ "target": "trn1",
45
+ "torch_dtype": "bfloat16",
46
+ "tp_degree": 2
47
+ },
48
+ "num_attention_heads": 32,
49
+ "num_hidden_layers": 40,
50
+ "num_key_value_heads": 8,
51
+ "residual_multiplier": 0.22,
52
+ "rms_norm_eps": 1e-05,
53
+ "rope_scaling": null,
54
+ "rope_theta": 5000000.0,
55
+ "tie_word_embeddings": true,
56
+ "use_cache": true,
57
+ "vocab_size": 49155
58
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/0cc25526e2cfc37a8875a3752f33c4d7505d8a07b869d0f3f41915cf6e763b74/9bb0e82ff320f6e2c0bc.json ADDED
@@ -0,0 +1,58 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "ibm-granite/granite-3.1-2b-instruct",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "GraniteForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.1,
10
+ "attention_multiplier": 0.015625,
11
+ "dtype": "bfloat16",
12
+ "embedding_multiplier": 12.0,
13
+ "hidden_act": "silu",
14
+ "hidden_size": 2048,
15
+ "initializer_range": 0.02,
16
+ "intermediate_size": 8192,
17
+ "logits_scaling": 8.0,
18
+ "max_position_embeddings": 131072,
19
+ "mlp_bias": false,
20
+ "model_type": "granite",
21
+ "neuron": {
22
+ "_serialized_key": "NxDNeuronConfig",
23
+ "batch_size": 1,
24
+ "capacity_factor": null,
25
+ "checkpoint_id": "ibm-granite/granite-3.1-2b-instruct",
26
+ "checkpoint_revision": "bbc2aed595bd38bd770263dc3ab831db9794441d",
27
+ "continuous_batching": false,
28
+ "ep_degree": 1,
29
+ "fused_qkv": true,
30
+ "glu_mlp": true,
31
+ "local_ranks_size": 2,
32
+ "max_batch_size": 1,
33
+ "max_context_length": 8192,
34
+ "max_topk": 256,
35
+ "n_active_tokens": 8192,
36
+ "neuronxcc_version": "2.21.33363.0+82129205",
37
+ "on_device_sampling": true,
38
+ "optimum_neuron_version": "0.4.5.dev1",
39
+ "output_logits": false,
40
+ "pp_degree": 1,
41
+ "sequence_length": 8192,
42
+ "speculation_length": 0,
43
+ "start_rank_id": 0,
44
+ "target": "trn1",
45
+ "torch_dtype": "bfloat16",
46
+ "tp_degree": 2
47
+ },
48
+ "num_attention_heads": 32,
49
+ "num_hidden_layers": 40,
50
+ "num_key_value_heads": 8,
51
+ "residual_multiplier": 0.22,
52
+ "rms_norm_eps": 1e-05,
53
+ "rope_scaling": null,
54
+ "rope_theta": 5000000.0,
55
+ "tie_word_embeddings": true,
56
+ "use_cache": true,
57
+ "vocab_size": 49155
58
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/2da7a00f0478d50ae1e7f75f085c5b2773b5f355f427c61cf34cb6febd629d96/4925f3a03aeddc47881d.json ADDED
@@ -0,0 +1,59 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "yujiepan/phi-4-tiny-random",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "Phi3ForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "auto_map": {},
11
+ "dtype": "bfloat16",
12
+ "embd_pdrop": 0.0,
13
+ "hidden_act": "silu",
14
+ "hidden_size": 16,
15
+ "initializer_range": 0.02,
16
+ "intermediate_size": 32,
17
+ "max_position_embeddings": 16384,
18
+ "model_type": "phi3",
19
+ "neuron": {
20
+ "_serialized_key": "NxDNeuronConfig",
21
+ "batch_size": 1,
22
+ "capacity_factor": null,
23
+ "checkpoint_id": "yujiepan/phi-4-tiny-random",
24
+ "checkpoint_revision": "18a9a1168dc97ac6d128f811925670c275610f5a",
25
+ "continuous_batching": false,
26
+ "ep_degree": 1,
27
+ "fused_qkv": true,
28
+ "glu_mlp": true,
29
+ "local_ranks_size": 2,
30
+ "max_batch_size": 1,
31
+ "max_context_length": 1024,
32
+ "max_topk": 256,
33
+ "n_active_tokens": 1024,
34
+ "neuronxcc_version": "2.21.33363.0+82129205",
35
+ "on_device_sampling": true,
36
+ "optimum_neuron_version": "0.4.5.dev1",
37
+ "output_logits": false,
38
+ "pp_degree": 1,
39
+ "sequence_length": 1024,
40
+ "speculation_length": 0,
41
+ "start_rank_id": 0,
42
+ "target": "trn1",
43
+ "torch_dtype": "bfloat16",
44
+ "tp_degree": 2
45
+ },
46
+ "num_attention_heads": 2,
47
+ "num_hidden_layers": 2,
48
+ "num_key_value_heads": 1,
49
+ "original_max_position_embeddings": 16384,
50
+ "partial_rotary_factor": 1.0,
51
+ "resid_pdrop": 0.0,
52
+ "rms_norm_eps": 1e-05,
53
+ "rope_scaling": null,
54
+ "rope_theta": 250000,
55
+ "sliding_window": null,
56
+ "tie_word_embeddings": false,
57
+ "use_cache": true,
58
+ "vocab_size": 100352
59
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/4ab8140bc7eb4a553d95855c5c2be2cf8c0fbab21b823d76183b6f51e98b6fc5/93ac0e7f7bd685a9d58b.json ADDED
@@ -0,0 +1,58 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "dacorvo/Mixtral-tiny",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "MixtralForCausalLM"
7
+ ],
8
+ "attention_dropout": 0.0,
9
+ "dtype": "float16",
10
+ "head_dim": 32,
11
+ "hidden_act": "silu",
12
+ "hidden_size": 1024,
13
+ "initializer_range": 0.02,
14
+ "intermediate_size": 3584,
15
+ "max_position_embeddings": 1024,
16
+ "model_type": "mixtral",
17
+ "neuron": {
18
+ "_serialized_key": "NxDNeuronConfig",
19
+ "batch_size": 1,
20
+ "capacity_factor": null,
21
+ "checkpoint_id": "dacorvo/Mixtral-tiny",
22
+ "checkpoint_revision": "c557ba205ddff6ea911f4719e0d543d6c08356b6",
23
+ "continuous_batching": false,
24
+ "ep_degree": 1,
25
+ "fused_qkv": false,
26
+ "glu_mlp": true,
27
+ "local_ranks_size": 2,
28
+ "max_batch_size": 1,
29
+ "max_context_length": 1024,
30
+ "max_topk": 256,
31
+ "n_active_tokens": 1024,
32
+ "neuronxcc_version": "2.21.33363.0+82129205",
33
+ "on_device_sampling": false,
34
+ "optimum_neuron_version": "0.4.5.dev1",
35
+ "output_logits": false,
36
+ "pp_degree": 1,
37
+ "sequence_length": 1024,
38
+ "speculation_length": 0,
39
+ "start_rank_id": 0,
40
+ "target": "trn1",
41
+ "torch_dtype": "float16",
42
+ "tp_degree": 2
43
+ },
44
+ "num_attention_heads": 32,
45
+ "num_experts_per_tok": 2,
46
+ "num_hidden_layers": 2,
47
+ "num_key_value_heads": 8,
48
+ "num_local_experts": 8,
49
+ "output_router_logits": false,
50
+ "rms_norm_eps": 1e-05,
51
+ "rope_theta": 10000.0,
52
+ "router_aux_loss_coef": 0.001,
53
+ "router_jitter_noise": 0.0,
54
+ "sliding_window": 4096,
55
+ "tie_word_embeddings": false,
56
+ "use_cache": true,
57
+ "vocab_size": 32000
58
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/4cb7aff9e2a15c151396f2b684013e39d6739f0dec83e5c9dabbfe9d5fcf77b7/34454e3efb576d636921.json ADDED
@@ -0,0 +1,82 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "Qwen/Qwen2.5-0.5B",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "Qwen2ForCausalLM"
7
+ ],
8
+ "attention_dropout": 0.0,
9
+ "dtype": "bfloat16",
10
+ "hidden_act": "silu",
11
+ "hidden_size": 896,
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 4864,
14
+ "layer_types": [
15
+ "full_attention",
16
+ "full_attention",
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
+ ],
40
+ "max_position_embeddings": 32768,
41
+ "max_window_layers": 24,
42
+ "model_type": "qwen2",
43
+ "neuron": {
44
+ "_serialized_key": "NxDNeuronConfig",
45
+ "batch_size": 1,
46
+ "capacity_factor": null,
47
+ "checkpoint_id": "Qwen/Qwen2.5-0.5B",
48
+ "checkpoint_revision": "060db6499f32faf8b98477b0a26969ef7d8b9987",
49
+ "continuous_batching": false,
50
+ "ep_degree": 1,
51
+ "fused_qkv": false,
52
+ "glu_mlp": true,
53
+ "local_ranks_size": 2,
54
+ "max_batch_size": 1,
55
+ "max_context_length": 8192,
56
+ "max_topk": 256,
57
+ "n_active_tokens": 8192,
58
+ "neuronxcc_version": "2.21.33363.0+82129205",
59
+ "on_device_sampling": true,
60
+ "optimum_neuron_version": "0.4.5.dev1",
61
+ "output_logits": false,
62
+ "pp_degree": 1,
63
+ "sequence_length": 8192,
64
+ "speculation_length": 0,
65
+ "start_rank_id": 0,
66
+ "target": "trn1",
67
+ "torch_dtype": "bfloat16",
68
+ "tp_degree": 2
69
+ },
70
+ "num_attention_heads": 14,
71
+ "num_hidden_layers": 24,
72
+ "num_key_value_heads": 2,
73
+ "rms_norm_eps": 1e-06,
74
+ "rope_scaling": null,
75
+ "rope_theta": 1000000.0,
76
+ "sliding_window": null,
77
+ "tie_word_embeddings": true,
78
+ "use_cache": true,
79
+ "use_mrope": false,
80
+ "use_sliding_window": false,
81
+ "vocab_size": 151936
82
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/4cb7aff9e2a15c151396f2b684013e39d6739f0dec83e5c9dabbfe9d5fcf77b7/5ec0bcb03ab4e2ad844c.json ADDED
@@ -0,0 +1,82 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "Qwen/Qwen2.5-0.5B",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "Qwen2ForCausalLM"
7
+ ],
8
+ "attention_dropout": 0.0,
9
+ "dtype": "bfloat16",
10
+ "hidden_act": "silu",
11
+ "hidden_size": 896,
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 4864,
14
+ "layer_types": [
15
+ "full_attention",
16
+ "full_attention",
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
+ ],
40
+ "max_position_embeddings": 32768,
41
+ "max_window_layers": 24,
42
+ "model_type": "qwen2",
43
+ "neuron": {
44
+ "_serialized_key": "NxDNeuronConfig",
45
+ "batch_size": 4,
46
+ "capacity_factor": null,
47
+ "checkpoint_id": "Qwen/Qwen2.5-0.5B",
48
+ "checkpoint_revision": "060db6499f32faf8b98477b0a26969ef7d8b9987",
49
+ "continuous_batching": true,
50
+ "ep_degree": 1,
51
+ "fused_qkv": false,
52
+ "glu_mlp": true,
53
+ "local_ranks_size": 2,
54
+ "max_batch_size": 4,
55
+ "max_context_length": 4096,
56
+ "max_topk": 256,
57
+ "n_active_tokens": 4096,
58
+ "neuronxcc_version": "2.21.33363.0+82129205",
59
+ "on_device_sampling": false,
60
+ "optimum_neuron_version": "0.4.5.dev1",
61
+ "output_logits": false,
62
+ "pp_degree": 1,
63
+ "sequence_length": 4096,
64
+ "speculation_length": 0,
65
+ "start_rank_id": 0,
66
+ "target": "trn1",
67
+ "torch_dtype": "bfloat16",
68
+ "tp_degree": 2
69
+ },
70
+ "num_attention_heads": 14,
71
+ "num_hidden_layers": 24,
72
+ "num_key_value_heads": 2,
73
+ "rms_norm_eps": 1e-06,
74
+ "rope_scaling": null,
75
+ "rope_theta": 1000000.0,
76
+ "sliding_window": null,
77
+ "tie_word_embeddings": true,
78
+ "use_cache": true,
79
+ "use_mrope": false,
80
+ "use_sliding_window": false,
81
+ "vocab_size": 151936
82
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/4cb7aff9e2a15c151396f2b684013e39d6739f0dec83e5c9dabbfe9d5fcf77b7/df54a6f9b56ca3bdfc41.json ADDED
@@ -0,0 +1,82 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "Qwen/Qwen2.5-0.5B",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "Qwen2ForCausalLM"
7
+ ],
8
+ "attention_dropout": 0.0,
9
+ "dtype": "bfloat16",
10
+ "hidden_act": "silu",
11
+ "hidden_size": 896,
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 4864,
14
+ "layer_types": [
15
+ "full_attention",
16
+ "full_attention",
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
+ ],
40
+ "max_position_embeddings": 32768,
41
+ "max_window_layers": 24,
42
+ "model_type": "qwen2",
43
+ "neuron": {
44
+ "_serialized_key": "NxDNeuronConfig",
45
+ "batch_size": 2,
46
+ "capacity_factor": null,
47
+ "checkpoint_id": "Qwen/Qwen2.5-0.5B",
48
+ "checkpoint_revision": "060db6499f32faf8b98477b0a26969ef7d8b9987",
49
+ "continuous_batching": true,
50
+ "ep_degree": 1,
51
+ "fused_qkv": false,
52
+ "glu_mlp": true,
53
+ "local_ranks_size": 2,
54
+ "max_batch_size": 2,
55
+ "max_context_length": 1024,
56
+ "max_topk": 256,
57
+ "n_active_tokens": 1024,
58
+ "neuronxcc_version": "2.21.33363.0+82129205",
59
+ "on_device_sampling": false,
60
+ "optimum_neuron_version": "0.4.5.dev1",
61
+ "output_logits": false,
62
+ "pp_degree": 1,
63
+ "sequence_length": 1024,
64
+ "speculation_length": 0,
65
+ "start_rank_id": 0,
66
+ "target": "trn1",
67
+ "torch_dtype": "bfloat16",
68
+ "tp_degree": 2
69
+ },
70
+ "num_attention_heads": 14,
71
+ "num_hidden_layers": 24,
72
+ "num_key_value_heads": 2,
73
+ "rms_norm_eps": 1e-06,
74
+ "rope_scaling": null,
75
+ "rope_theta": 1000000.0,
76
+ "sliding_window": null,
77
+ "tie_word_embeddings": true,
78
+ "use_cache": true,
79
+ "use_mrope": false,
80
+ "use_sliding_window": false,
81
+ "vocab_size": 151936
82
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/4d281af1651143f9a83b42af25d9f0a533dbbd9c3441537127ae772fd3e399cc/2c2bad80f80ab55a2d84.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "Qwen/Qwen3-Embedding-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": 32768,
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-Embedding-0.6B",
54
+ "checkpoint_revision": "c54f2e6e80b2d7b7de06f51cec4959f6b3e03418",
55
+ "continuous_batching": false,
56
+ "ep_degree": 1,
57
+ "fused_qkv": true,
58
+ "glu_mlp": true,
59
+ "local_ranks_size": 2,
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": false,
66
+ "optimum_neuron_version": "0.4.5.dev1",
67
+ "output_logits": false,
68
+ "pp_degree": 1,
69
+ "sequence_length": 1024,
70
+ "speculation_length": 0,
71
+ "start_rank_id": 0,
72
+ "target": "trn1",
73
+ "torch_dtype": "bfloat16",
74
+ "tp_degree": 2
75
+ },
76
+ "num_attention_heads": 16,
77
+ "num_hidden_layers": 28,
78
+ "num_key_value_heads": 8,
79
+ "rms_norm_eps": 1e-06,
80
+ "rope_scaling": null,
81
+ "rope_theta": 1000000,
82
+ "sliding_window": null,
83
+ "tie_word_embeddings": true,
84
+ "use_cache": true,
85
+ "use_sliding_window": false,
86
+ "vocab_size": 151669
87
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/4d281af1651143f9a83b42af25d9f0a533dbbd9c3441537127ae772fd3e399cc/9c8cb869b5d7e2540fa0.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "Qwen/Qwen3-Embedding-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": 32768,
47
+ "max_window_layers": 28,
48
+ "model_type": "qwen3",
49
+ "neuron": {
50
+ "_serialized_key": "NxDNeuronConfig",
51
+ "batch_size": 6,
52
+ "capacity_factor": null,
53
+ "checkpoint_id": "Qwen/Qwen3-Embedding-0.6B",
54
+ "checkpoint_revision": "c54f2e6e80b2d7b7de06f51cec4959f6b3e03418",
55
+ "continuous_batching": false,
56
+ "ep_degree": 1,
57
+ "fused_qkv": true,
58
+ "glu_mlp": true,
59
+ "local_ranks_size": 2,
60
+ "max_batch_size": 6,
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": false,
66
+ "optimum_neuron_version": "0.4.5.dev1",
67
+ "output_logits": false,
68
+ "pp_degree": 1,
69
+ "sequence_length": 1024,
70
+ "speculation_length": 0,
71
+ "start_rank_id": 0,
72
+ "target": "trn1",
73
+ "torch_dtype": "bfloat16",
74
+ "tp_degree": 2
75
+ },
76
+ "num_attention_heads": 16,
77
+ "num_hidden_layers": 28,
78
+ "num_key_value_heads": 8,
79
+ "rms_norm_eps": 1e-06,
80
+ "rope_scaling": null,
81
+ "rope_theta": 1000000,
82
+ "sliding_window": null,
83
+ "tie_word_embeddings": true,
84
+ "use_cache": true,
85
+ "use_sliding_window": false,
86
+ "vocab_size": 151669
87
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/6454afdf3e9d66c7226c13a575b718845c25e53b0699600ba2bb4f883e9d841b/12ce6ef4377bb2672028.json ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "llamafactory/tiny-random-Llama-3",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "LlamaForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "dtype": "float16",
11
+ "head_dim": 4,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 16,
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 64,
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": "llamafactory/tiny-random-Llama-3",
24
+ "checkpoint_revision": "bf2a2e3bf199ad2ee96f02a3c00246c608db22a8",
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": 128,
32
+ "max_topk": 256,
33
+ "n_active_tokens": 128,
34
+ "neuronxcc_version": "2.21.33363.0+82129205",
35
+ "on_device_sampling": true,
36
+ "optimum_neuron_version": "0.4.5.dev1",
37
+ "output_logits": false,
38
+ "pp_degree": 1,
39
+ "sequence_length": 128,
40
+ "speculation_length": 0,
41
+ "start_rank_id": 0,
42
+ "target": "trn2",
43
+ "torch_dtype": "float16",
44
+ "tp_degree": 2
45
+ },
46
+ "num_attention_heads": 4,
47
+ "num_hidden_layers": 2,
48
+ "num_key_value_heads": 4,
49
+ "pretraining_tp": 1,
50
+ "rms_norm_eps": 1e-05,
51
+ "rope_scaling": {
52
+ "factor": 8.0,
53
+ "high_freq_factor": 4.0,
54
+ "low_freq_factor": 1.0,
55
+ "original_max_position_embeddings": 8192,
56
+ "rope_type": "llama3"
57
+ },
58
+ "rope_theta": 500000.0,
59
+ "tie_word_embeddings": false,
60
+ "use_cache": true,
61
+ "vocab_size": 128256
62
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/6454afdf3e9d66c7226c13a575b718845c25e53b0699600ba2bb4f883e9d841b/5997ab86579aadcc5cdd.json ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "llamafactory/tiny-random-Llama-3",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "LlamaForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "dtype": "float16",
11
+ "head_dim": 4,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 16,
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 64,
16
+ "max_position_embeddings": 131072,
17
+ "mlp_bias": false,
18
+ "model_type": "llama",
19
+ "neuron": {
20
+ "_serialized_key": "NxDNeuronConfig",
21
+ "batch_size": 1,
22
+ "capacity_factor": null,
23
+ "checkpoint_id": "llamafactory/tiny-random-Llama-3",
24
+ "checkpoint_revision": "bf2a2e3bf199ad2ee96f02a3c00246c608db22a8",
25
+ "continuous_batching": false,
26
+ "ep_degree": 1,
27
+ "fused_qkv": true,
28
+ "glu_mlp": true,
29
+ "local_ranks_size": 2,
30
+ "max_batch_size": 1,
31
+ "max_context_length": 1024,
32
+ "max_topk": 256,
33
+ "n_active_tokens": 1024,
34
+ "neuronxcc_version": "2.21.33363.0+82129205",
35
+ "on_device_sampling": true,
36
+ "optimum_neuron_version": "0.4.5.dev1",
37
+ "output_logits": false,
38
+ "pp_degree": 1,
39
+ "sequence_length": 1024,
40
+ "speculation_length": 0,
41
+ "start_rank_id": 0,
42
+ "target": "trn1",
43
+ "torch_dtype": "float16",
44
+ "tp_degree": 2
45
+ },
46
+ "num_attention_heads": 4,
47
+ "num_hidden_layers": 2,
48
+ "num_key_value_heads": 4,
49
+ "pretraining_tp": 1,
50
+ "rms_norm_eps": 1e-05,
51
+ "rope_scaling": {
52
+ "factor": 8.0,
53
+ "high_freq_factor": 4.0,
54
+ "low_freq_factor": 1.0,
55
+ "original_max_position_embeddings": 8192,
56
+ "rope_type": "llama3"
57
+ },
58
+ "rope_theta": 500000.0,
59
+ "tie_word_embeddings": false,
60
+ "use_cache": true,
61
+ "vocab_size": 128256
62
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/6454afdf3e9d66c7226c13a575b718845c25e53b0699600ba2bb4f883e9d841b/f85a0d1bea066bc1274b.json ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "llamafactory/tiny-random-Llama-3",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "LlamaForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "dtype": "float16",
11
+ "head_dim": 4,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 16,
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 64,
16
+ "max_position_embeddings": 131072,
17
+ "mlp_bias": false,
18
+ "model_type": "llama",
19
+ "neuron": {
20
+ "_serialized_key": "NxDNeuronConfig",
21
+ "batch_size": 1,
22
+ "capacity_factor": null,
23
+ "checkpoint_id": "llamafactory/tiny-random-Llama-3",
24
+ "checkpoint_revision": "bf2a2e3bf199ad2ee96f02a3c00246c608db22a8",
25
+ "continuous_batching": false,
26
+ "ep_degree": 1,
27
+ "fused_qkv": true,
28
+ "glu_mlp": true,
29
+ "local_ranks_size": 2,
30
+ "max_batch_size": 1,
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": "0.4.5.dev1",
37
+ "output_logits": false,
38
+ "pp_degree": 1,
39
+ "sequence_length": 512,
40
+ "speculation_length": 0,
41
+ "start_rank_id": 0,
42
+ "target": "trn1",
43
+ "torch_dtype": "float16",
44
+ "tp_degree": 2
45
+ },
46
+ "num_attention_heads": 4,
47
+ "num_hidden_layers": 2,
48
+ "num_key_value_heads": 4,
49
+ "pretraining_tp": 1,
50
+ "rms_norm_eps": 1e-05,
51
+ "rope_scaling": {
52
+ "factor": 8.0,
53
+ "high_freq_factor": 4.0,
54
+ "low_freq_factor": 1.0,
55
+ "original_max_position_embeddings": 8192,
56
+ "rope_type": "llama3"
57
+ },
58
+ "rope_theta": 500000.0,
59
+ "tie_word_embeddings": false,
60
+ "use_cache": true,
61
+ "vocab_size": 128256
62
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/73707b485eab9008c7aba7f5dad0ce2384ac685318d5f888c12fa0d81ed90b19/1a86e1cfcaf229143d5c.json ADDED
@@ -0,0 +1,134 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "HuggingFaceTB/SmolLM3-3B",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "SmolLM3ForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "dtype": "bfloat16",
11
+ "hidden_act": "silu",
12
+ "hidden_size": 2048,
13
+ "initializer_range": 0.02,
14
+ "intermediate_size": 11008,
15
+ "layer_types": [
16
+ "full_attention",
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
+ ],
53
+ "max_position_embeddings": 65536,
54
+ "max_window_layers": 28,
55
+ "mlp_bias": false,
56
+ "model_type": "smollm3",
57
+ "neuron": {
58
+ "_serialized_key": "NxDNeuronConfig",
59
+ "batch_size": 1,
60
+ "capacity_factor": null,
61
+ "checkpoint_id": "HuggingFaceTB/SmolLM3-3B",
62
+ "checkpoint_revision": "a07cc9a04f16550a088caea529712d1d335b0ac1",
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": true,
74
+ "optimum_neuron_version": "0.4.5.dev1",
75
+ "output_logits": false,
76
+ "pp_degree": 1,
77
+ "sequence_length": 1024,
78
+ "speculation_length": 0,
79
+ "start_rank_id": 0,
80
+ "target": "trn1",
81
+ "torch_dtype": "bfloat16",
82
+ "tp_degree": 2
83
+ },
84
+ "no_rope_layer_interval": 4,
85
+ "no_rope_layers": [
86
+ 1,
87
+ 1,
88
+ 1,
89
+ 0,
90
+ 1,
91
+ 1,
92
+ 1,
93
+ 0,
94
+ 1,
95
+ 1,
96
+ 1,
97
+ 0,
98
+ 1,
99
+ 1,
100
+ 1,
101
+ 0,
102
+ 1,
103
+ 1,
104
+ 1,
105
+ 0,
106
+ 1,
107
+ 1,
108
+ 1,
109
+ 0,
110
+ 1,
111
+ 1,
112
+ 1,
113
+ 0,
114
+ 1,
115
+ 1,
116
+ 1,
117
+ 0,
118
+ 1,
119
+ 1,
120
+ 1,
121
+ 0
122
+ ],
123
+ "num_attention_heads": 16,
124
+ "num_hidden_layers": 36,
125
+ "num_key_value_heads": 4,
126
+ "pretraining_tp": 2,
127
+ "rms_norm_eps": 1e-06,
128
+ "rope_scaling": null,
129
+ "rope_theta": 5000000.0,
130
+ "sliding_window": null,
131
+ "use_cache": false,
132
+ "use_sliding_window": false,
133
+ "vocab_size": 128256
134
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/73707b485eab9008c7aba7f5dad0ce2384ac685318d5f888c12fa0d81ed90b19/7144c43faca4fea18b01.json ADDED
@@ -0,0 +1,134 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "HuggingFaceTB/SmolLM3-3B",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "SmolLM3ForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "dtype": "bfloat16",
11
+ "hidden_act": "silu",
12
+ "hidden_size": 2048,
13
+ "initializer_range": 0.02,
14
+ "intermediate_size": 11008,
15
+ "layer_types": [
16
+ "full_attention",
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
+ ],
53
+ "max_position_embeddings": 65536,
54
+ "max_window_layers": 28,
55
+ "mlp_bias": false,
56
+ "model_type": "smollm3",
57
+ "neuron": {
58
+ "_serialized_key": "NxDNeuronConfig",
59
+ "batch_size": 4,
60
+ "capacity_factor": null,
61
+ "checkpoint_id": "HuggingFaceTB/SmolLM3-3B",
62
+ "checkpoint_revision": "a07cc9a04f16550a088caea529712d1d335b0ac1",
63
+ "continuous_batching": true,
64
+ "ep_degree": 1,
65
+ "fused_qkv": true,
66
+ "glu_mlp": true,
67
+ "local_ranks_size": 2,
68
+ "max_batch_size": 4,
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": true,
74
+ "optimum_neuron_version": "0.4.5.dev1",
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
+ "no_rope_layer_interval": 4,
85
+ "no_rope_layers": [
86
+ 1,
87
+ 1,
88
+ 1,
89
+ 0,
90
+ 1,
91
+ 1,
92
+ 1,
93
+ 0,
94
+ 1,
95
+ 1,
96
+ 1,
97
+ 0,
98
+ 1,
99
+ 1,
100
+ 1,
101
+ 0,
102
+ 1,
103
+ 1,
104
+ 1,
105
+ 0,
106
+ 1,
107
+ 1,
108
+ 1,
109
+ 0,
110
+ 1,
111
+ 1,
112
+ 1,
113
+ 0,
114
+ 1,
115
+ 1,
116
+ 1,
117
+ 0,
118
+ 1,
119
+ 1,
120
+ 1,
121
+ 0
122
+ ],
123
+ "num_attention_heads": 16,
124
+ "num_hidden_layers": 36,
125
+ "num_key_value_heads": 4,
126
+ "pretraining_tp": 2,
127
+ "rms_norm_eps": 1e-06,
128
+ "rope_scaling": null,
129
+ "rope_theta": 5000000.0,
130
+ "sliding_window": null,
131
+ "use_cache": false,
132
+ "use_sliding_window": false,
133
+ "vocab_size": 128256
134
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/73707b485eab9008c7aba7f5dad0ce2384ac685318d5f888c12fa0d81ed90b19/eb6ead215819eab7b04a.json ADDED
@@ -0,0 +1,134 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "HuggingFaceTB/SmolLM3-3B",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "SmolLM3ForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "dtype": "bfloat16",
11
+ "hidden_act": "silu",
12
+ "hidden_size": 2048,
13
+ "initializer_range": 0.02,
14
+ "intermediate_size": 11008,
15
+ "layer_types": [
16
+ "full_attention",
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
+ ],
53
+ "max_position_embeddings": 65536,
54
+ "max_window_layers": 28,
55
+ "mlp_bias": false,
56
+ "model_type": "smollm3",
57
+ "neuron": {
58
+ "_serialized_key": "NxDNeuronConfig",
59
+ "batch_size": 1,
60
+ "capacity_factor": null,
61
+ "checkpoint_id": "HuggingFaceTB/SmolLM3-3B",
62
+ "checkpoint_revision": "a07cc9a04f16550a088caea529712d1d335b0ac1",
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": true,
74
+ "optimum_neuron_version": "0.4.5.dev1",
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
+ "no_rope_layer_interval": 4,
85
+ "no_rope_layers": [
86
+ 1,
87
+ 1,
88
+ 1,
89
+ 0,
90
+ 1,
91
+ 1,
92
+ 1,
93
+ 0,
94
+ 1,
95
+ 1,
96
+ 1,
97
+ 0,
98
+ 1,
99
+ 1,
100
+ 1,
101
+ 0,
102
+ 1,
103
+ 1,
104
+ 1,
105
+ 0,
106
+ 1,
107
+ 1,
108
+ 1,
109
+ 0,
110
+ 1,
111
+ 1,
112
+ 1,
113
+ 0,
114
+ 1,
115
+ 1,
116
+ 1,
117
+ 0,
118
+ 1,
119
+ 1,
120
+ 1,
121
+ 0
122
+ ],
123
+ "num_attention_heads": 16,
124
+ "num_hidden_layers": 36,
125
+ "num_key_value_heads": 4,
126
+ "pretraining_tp": 2,
127
+ "rms_norm_eps": 1e-06,
128
+ "rope_scaling": null,
129
+ "rope_theta": 5000000.0,
130
+ "sliding_window": null,
131
+ "use_cache": false,
132
+ "use_sliding_window": false,
133
+ "vocab_size": 128256
134
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/7518518c7e077820070186deda960d8cc49db068cdf0ac70664098fa2b6b698c/a037d3d5582033b493b7.json ADDED
@@ -0,0 +1,64 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "yujiepan/qwen2.5-128k-tiny-random",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "Qwen2ForCausalLM"
7
+ ],
8
+ "attention_dropout": 0.0,
9
+ "dtype": "bfloat16",
10
+ "hidden_act": "silu",
11
+ "hidden_size": 8,
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 16,
14
+ "layer_types": [
15
+ "full_attention",
16
+ "full_attention"
17
+ ],
18
+ "max_position_embeddings": 32768,
19
+ "max_window_layers": 1,
20
+ "model_type": "qwen2",
21
+ "neuron": {
22
+ "_serialized_key": "NxDNeuronConfig",
23
+ "batch_size": 1,
24
+ "capacity_factor": null,
25
+ "checkpoint_id": "yujiepan/qwen2.5-128k-tiny-random",
26
+ "checkpoint_revision": "c8296d4ca3f87782876d2382fbb6481d1beb8ef0",
27
+ "continuous_batching": false,
28
+ "ep_degree": 1,
29
+ "fused_qkv": false,
30
+ "glu_mlp": true,
31
+ "local_ranks_size": 2,
32
+ "max_batch_size": 1,
33
+ "max_context_length": 1024,
34
+ "max_topk": 256,
35
+ "n_active_tokens": 1024,
36
+ "neuronxcc_version": "2.21.33363.0+82129205",
37
+ "on_device_sampling": true,
38
+ "optimum_neuron_version": "0.4.5.dev1",
39
+ "output_logits": false,
40
+ "pp_degree": 1,
41
+ "sequence_length": 1024,
42
+ "speculation_length": 0,
43
+ "start_rank_id": 0,
44
+ "target": "trn1",
45
+ "torch_dtype": "bfloat16",
46
+ "tp_degree": 2
47
+ },
48
+ "num_attention_heads": 4,
49
+ "num_hidden_layers": 2,
50
+ "num_key_value_heads": 2,
51
+ "rms_norm_eps": 1e-06,
52
+ "rope_scaling": {
53
+ "factor": 4.0,
54
+ "original_max_position_embeddings": 32768,
55
+ "rope_type": "yarn",
56
+ "type": "yarn"
57
+ },
58
+ "rope_theta": 1000000.0,
59
+ "sliding_window": null,
60
+ "tie_word_embeddings": false,
61
+ "use_cache": true,
62
+ "use_sliding_window": false,
63
+ "vocab_size": 152064
64
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/8c90ac2593ed0b7f1ecb60e82cb184fb11f2ea640befa1cc7b10766a5c02525d/8cefa5ff9083418aa040.json ADDED
@@ -0,0 +1,164 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "microsoft/Phi-3.5-mini-instruct",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "Phi3ForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "auto_map": {
11
+ "AutoConfig": "configuration_phi3.Phi3Config",
12
+ "AutoModelForCausalLM": "modeling_phi3.Phi3ForCausalLM"
13
+ },
14
+ "dtype": "bfloat16",
15
+ "embd_pdrop": 0.0,
16
+ "hidden_act": "silu",
17
+ "hidden_size": 3072,
18
+ "initializer_range": 0.02,
19
+ "intermediate_size": 8192,
20
+ "max_position_embeddings": 131072,
21
+ "model_type": "phi3",
22
+ "neuron": {
23
+ "_serialized_key": "NxDNeuronConfig",
24
+ "batch_size": 4,
25
+ "capacity_factor": null,
26
+ "checkpoint_id": "microsoft/Phi-3.5-mini-instruct",
27
+ "checkpoint_revision": "2fe192450127e6a83f7441aef6e3ca586c338b77",
28
+ "continuous_batching": true,
29
+ "ep_degree": 1,
30
+ "fused_qkv": true,
31
+ "glu_mlp": true,
32
+ "local_ranks_size": 2,
33
+ "max_batch_size": 4,
34
+ "max_context_length": 4096,
35
+ "max_topk": 256,
36
+ "n_active_tokens": 4096,
37
+ "neuronxcc_version": "2.21.33363.0+82129205",
38
+ "on_device_sampling": true,
39
+ "optimum_neuron_version": "0.4.5.dev1",
40
+ "output_logits": false,
41
+ "pp_degree": 1,
42
+ "sequence_length": 4096,
43
+ "speculation_length": 0,
44
+ "start_rank_id": 0,
45
+ "target": "trn1",
46
+ "torch_dtype": "bfloat16",
47
+ "tp_degree": 2
48
+ },
49
+ "num_attention_heads": 32,
50
+ "num_hidden_layers": 32,
51
+ "num_key_value_heads": 32,
52
+ "original_max_position_embeddings": 4096,
53
+ "partial_rotary_factor": 1.0,
54
+ "resid_pdrop": 0.0,
55
+ "rms_norm_eps": 1e-05,
56
+ "rope_scaling": {
57
+ "long_factor": [
58
+ 1.0800000429153442,
59
+ 1.1100000143051147,
60
+ 1.1399999856948853,
61
+ 1.340000033378601,
62
+ 1.5899999141693115,
63
+ 1.600000023841858,
64
+ 1.6200000047683716,
65
+ 2.620000123977661,
66
+ 3.2300000190734863,
67
+ 3.2300000190734863,
68
+ 4.789999961853027,
69
+ 7.400000095367432,
70
+ 7.700000286102295,
71
+ 9.09000015258789,
72
+ 12.199999809265137,
73
+ 17.670000076293945,
74
+ 24.46000099182129,
75
+ 28.57000160217285,
76
+ 30.420001983642578,
77
+ 30.840002059936523,
78
+ 32.590003967285156,
79
+ 32.93000411987305,
80
+ 42.320003509521484,
81
+ 44.96000289916992,
82
+ 50.340003967285156,
83
+ 50.45000457763672,
84
+ 57.55000305175781,
85
+ 57.93000411987305,
86
+ 58.21000289916992,
87
+ 60.1400032043457,
88
+ 62.61000442504883,
89
+ 62.62000274658203,
90
+ 62.71000289916992,
91
+ 63.1400032043457,
92
+ 63.1400032043457,
93
+ 63.77000427246094,
94
+ 63.93000411987305,
95
+ 63.96000289916992,
96
+ 63.970001220703125,
97
+ 64.02999877929688,
98
+ 64.06999969482422,
99
+ 64.08000183105469,
100
+ 64.12000274658203,
101
+ 64.41000366210938,
102
+ 64.4800033569336,
103
+ 64.51000213623047,
104
+ 64.52999877929688,
105
+ 64.83999633789062
106
+ ],
107
+ "short_factor": [
108
+ 1.0,
109
+ 1.0199999809265137,
110
+ 1.0299999713897705,
111
+ 1.0299999713897705,
112
+ 1.0499999523162842,
113
+ 1.0499999523162842,
114
+ 1.0499999523162842,
115
+ 1.0499999523162842,
116
+ 1.0499999523162842,
117
+ 1.0699999332427979,
118
+ 1.0999999046325684,
119
+ 1.1099998950958252,
120
+ 1.1599998474121094,
121
+ 1.1599998474121094,
122
+ 1.1699998378753662,
123
+ 1.2899998426437378,
124
+ 1.339999794960022,
125
+ 1.679999828338623,
126
+ 1.7899998426437378,
127
+ 1.8199998140335083,
128
+ 1.8499997854232788,
129
+ 1.8799997568130493,
130
+ 1.9099997282028198,
131
+ 1.9399996995925903,
132
+ 1.9899996519088745,
133
+ 2.0199997425079346,
134
+ 2.0199997425079346,
135
+ 2.0199997425079346,
136
+ 2.0199997425079346,
137
+ 2.0199997425079346,
138
+ 2.0199997425079346,
139
+ 2.0299997329711914,
140
+ 2.0299997329711914,
141
+ 2.0299997329711914,
142
+ 2.0299997329711914,
143
+ 2.0299997329711914,
144
+ 2.0299997329711914,
145
+ 2.0299997329711914,
146
+ 2.0299997329711914,
147
+ 2.0299997329711914,
148
+ 2.0799996852874756,
149
+ 2.0899996757507324,
150
+ 2.189999580383301,
151
+ 2.2199995517730713,
152
+ 2.5899994373321533,
153
+ 2.729999542236328,
154
+ 2.749999523162842,
155
+ 2.8399994373321533
156
+ ],
157
+ "type": "longrope"
158
+ },
159
+ "rope_theta": 10000.0,
160
+ "sliding_window": 262144,
161
+ "tie_word_embeddings": false,
162
+ "use_cache": true,
163
+ "vocab_size": 32064
164
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/8c90ac2593ed0b7f1ecb60e82cb184fb11f2ea640befa1cc7b10766a5c02525d/9fbd66c92baabc8b5ad0.json ADDED
@@ -0,0 +1,164 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "microsoft/Phi-3.5-mini-instruct",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "Phi3ForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "auto_map": {
11
+ "AutoConfig": "configuration_phi3.Phi3Config",
12
+ "AutoModelForCausalLM": "modeling_phi3.Phi3ForCausalLM"
13
+ },
14
+ "dtype": "bfloat16",
15
+ "embd_pdrop": 0.0,
16
+ "hidden_act": "silu",
17
+ "hidden_size": 3072,
18
+ "initializer_range": 0.02,
19
+ "intermediate_size": 8192,
20
+ "max_position_embeddings": 131072,
21
+ "model_type": "phi3",
22
+ "neuron": {
23
+ "_serialized_key": "NxDNeuronConfig",
24
+ "batch_size": 1,
25
+ "capacity_factor": null,
26
+ "checkpoint_id": "microsoft/Phi-3.5-mini-instruct",
27
+ "checkpoint_revision": "2fe192450127e6a83f7441aef6e3ca586c338b77",
28
+ "continuous_batching": false,
29
+ "ep_degree": 1,
30
+ "fused_qkv": true,
31
+ "glu_mlp": true,
32
+ "local_ranks_size": 2,
33
+ "max_batch_size": 1,
34
+ "max_context_length": 8192,
35
+ "max_topk": 256,
36
+ "n_active_tokens": 8192,
37
+ "neuronxcc_version": "2.21.33363.0+82129205",
38
+ "on_device_sampling": true,
39
+ "optimum_neuron_version": "0.4.5.dev1",
40
+ "output_logits": false,
41
+ "pp_degree": 1,
42
+ "sequence_length": 8192,
43
+ "speculation_length": 0,
44
+ "start_rank_id": 0,
45
+ "target": "trn1",
46
+ "torch_dtype": "bfloat16",
47
+ "tp_degree": 2
48
+ },
49
+ "num_attention_heads": 32,
50
+ "num_hidden_layers": 32,
51
+ "num_key_value_heads": 32,
52
+ "original_max_position_embeddings": 4096,
53
+ "partial_rotary_factor": 1.0,
54
+ "resid_pdrop": 0.0,
55
+ "rms_norm_eps": 1e-05,
56
+ "rope_scaling": {
57
+ "long_factor": [
58
+ 1.0800000429153442,
59
+ 1.1100000143051147,
60
+ 1.1399999856948853,
61
+ 1.340000033378601,
62
+ 1.5899999141693115,
63
+ 1.600000023841858,
64
+ 1.6200000047683716,
65
+ 2.620000123977661,
66
+ 3.2300000190734863,
67
+ 3.2300000190734863,
68
+ 4.789999961853027,
69
+ 7.400000095367432,
70
+ 7.700000286102295,
71
+ 9.09000015258789,
72
+ 12.199999809265137,
73
+ 17.670000076293945,
74
+ 24.46000099182129,
75
+ 28.57000160217285,
76
+ 30.420001983642578,
77
+ 30.840002059936523,
78
+ 32.590003967285156,
79
+ 32.93000411987305,
80
+ 42.320003509521484,
81
+ 44.96000289916992,
82
+ 50.340003967285156,
83
+ 50.45000457763672,
84
+ 57.55000305175781,
85
+ 57.93000411987305,
86
+ 58.21000289916992,
87
+ 60.1400032043457,
88
+ 62.61000442504883,
89
+ 62.62000274658203,
90
+ 62.71000289916992,
91
+ 63.1400032043457,
92
+ 63.1400032043457,
93
+ 63.77000427246094,
94
+ 63.93000411987305,
95
+ 63.96000289916992,
96
+ 63.970001220703125,
97
+ 64.02999877929688,
98
+ 64.06999969482422,
99
+ 64.08000183105469,
100
+ 64.12000274658203,
101
+ 64.41000366210938,
102
+ 64.4800033569336,
103
+ 64.51000213623047,
104
+ 64.52999877929688,
105
+ 64.83999633789062
106
+ ],
107
+ "short_factor": [
108
+ 1.0,
109
+ 1.0199999809265137,
110
+ 1.0299999713897705,
111
+ 1.0299999713897705,
112
+ 1.0499999523162842,
113
+ 1.0499999523162842,
114
+ 1.0499999523162842,
115
+ 1.0499999523162842,
116
+ 1.0499999523162842,
117
+ 1.0699999332427979,
118
+ 1.0999999046325684,
119
+ 1.1099998950958252,
120
+ 1.1599998474121094,
121
+ 1.1599998474121094,
122
+ 1.1699998378753662,
123
+ 1.2899998426437378,
124
+ 1.339999794960022,
125
+ 1.679999828338623,
126
+ 1.7899998426437378,
127
+ 1.8199998140335083,
128
+ 1.8499997854232788,
129
+ 1.8799997568130493,
130
+ 1.9099997282028198,
131
+ 1.9399996995925903,
132
+ 1.9899996519088745,
133
+ 2.0199997425079346,
134
+ 2.0199997425079346,
135
+ 2.0199997425079346,
136
+ 2.0199997425079346,
137
+ 2.0199997425079346,
138
+ 2.0199997425079346,
139
+ 2.0299997329711914,
140
+ 2.0299997329711914,
141
+ 2.0299997329711914,
142
+ 2.0299997329711914,
143
+ 2.0299997329711914,
144
+ 2.0299997329711914,
145
+ 2.0299997329711914,
146
+ 2.0299997329711914,
147
+ 2.0299997329711914,
148
+ 2.0799996852874756,
149
+ 2.0899996757507324,
150
+ 2.189999580383301,
151
+ 2.2199995517730713,
152
+ 2.5899994373321533,
153
+ 2.729999542236328,
154
+ 2.749999523162842,
155
+ 2.8399994373321533
156
+ ],
157
+ "type": "longrope"
158
+ },
159
+ "rope_theta": 10000.0,
160
+ "sliding_window": 262144,
161
+ "tie_word_embeddings": false,
162
+ "use_cache": true,
163
+ "vocab_size": 32064
164
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/920f44ce6d3e004d1ce547ae06644f7be262180644b04573153aa15d98742edc/ada37737b07aadde24a9.json ADDED
@@ -0,0 +1,65 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "optimum-internal-testing/tiny-random-qwen3_moe",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "Qwen3MoeForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "decoder_sparse_step": 2,
11
+ "dtype": "float32",
12
+ "head_dim": 32,
13
+ "hidden_act": "silu",
14
+ "hidden_size": 64,
15
+ "initializer_range": 0.02,
16
+ "intermediate_size": 128,
17
+ "max_position_embeddings": 40960,
18
+ "max_window_layers": 1,
19
+ "mlp_only_layers": [],
20
+ "model_type": "qwen3_moe",
21
+ "moe_intermediate_size": 128,
22
+ "neuron": {
23
+ "_serialized_key": "NxDNeuronConfig",
24
+ "batch_size": 1,
25
+ "capacity_factor": null,
26
+ "checkpoint_id": "optimum-internal-testing/tiny-random-qwen3_moe",
27
+ "checkpoint_revision": "e0230be2839556b44b7400a233c73c74b4abb7af",
28
+ "continuous_batching": false,
29
+ "ep_degree": 1,
30
+ "fused_qkv": false,
31
+ "glu_mlp": true,
32
+ "local_ranks_size": 2,
33
+ "max_batch_size": 1,
34
+ "max_context_length": 1024,
35
+ "max_topk": 256,
36
+ "n_active_tokens": 1024,
37
+ "neuronxcc_version": "2.21.33363.0+82129205",
38
+ "on_device_sampling": true,
39
+ "optimum_neuron_version": "0.4.5.dev1",
40
+ "output_logits": false,
41
+ "pp_degree": 1,
42
+ "sequence_length": 1024,
43
+ "speculation_length": 0,
44
+ "start_rank_id": 0,
45
+ "target": "trn1",
46
+ "torch_dtype": "float32",
47
+ "tp_degree": 2
48
+ },
49
+ "norm_topk_prob": true,
50
+ "num_attention_heads": 2,
51
+ "num_experts": 8,
52
+ "num_experts_per_tok": 2,
53
+ "num_hidden_layers": 2,
54
+ "num_key_value_heads": 1,
55
+ "output_router_logits": false,
56
+ "rms_norm_eps": 1e-06,
57
+ "rope_scaling": null,
58
+ "rope_theta": 1000000.0,
59
+ "router_aux_loss_coef": 0.001,
60
+ "sliding_window": null,
61
+ "tie_word_embeddings": true,
62
+ "use_cache": true,
63
+ "use_sliding_window": false,
64
+ "vocab_size": 151936
65
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/929b02754a13cbfdf657d863c3fc6f3bce672879bc6ae48ab45be21e881e9ec2/97503f4e2dca15bd6721.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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": 1,
52
+ "capacity_factor": null,
53
+ "checkpoint_id": "Qwen/Qwen3-0.6B",
54
+ "checkpoint_revision": "c1899de289a04d12100db370d81485cdf75e47ca",
55
+ "continuous_batching": false,
56
+ "ep_degree": 1,
57
+ "fused_qkv": true,
58
+ "glu_mlp": true,
59
+ "local_ranks_size": 2,
60
+ "max_batch_size": 1,
61
+ "max_context_length": 8192,
62
+ "max_topk": 256,
63
+ "n_active_tokens": 8192,
64
+ "neuronxcc_version": "2.21.33363.0+82129205",
65
+ "on_device_sampling": true,
66
+ "optimum_neuron_version": "0.4.5.dev1",
67
+ "output_logits": false,
68
+ "pp_degree": 1,
69
+ "sequence_length": 8192,
70
+ "speculation_length": 0,
71
+ "start_rank_id": 0,
72
+ "target": "trn1",
73
+ "torch_dtype": "bfloat16",
74
+ "tp_degree": 2
75
+ },
76
+ "num_attention_heads": 16,
77
+ "num_hidden_layers": 28,
78
+ "num_key_value_heads": 8,
79
+ "rms_norm_eps": 1e-06,
80
+ "rope_scaling": null,
81
+ "rope_theta": 1000000,
82
+ "sliding_window": null,
83
+ "tie_word_embeddings": true,
84
+ "use_cache": true,
85
+ "use_sliding_window": false,
86
+ "vocab_size": 151936
87
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/9c24a2080ec26e9c74d5bd90866f74131aa4d5c975415b96f5e6600f081d5647/a7d7ccbe80fcc54f090d.json ADDED
@@ -0,0 +1,81 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "tiny-random/llama-4",
4
+ "_task": "text-generation",
5
+ "attention_bias": false,
6
+ "attention_chunk_size": 128,
7
+ "attention_dropout": 0.0,
8
+ "attn_scale": 0.1,
9
+ "attn_temperature_tuning": 4,
10
+ "cache_implementation": "hybrid",
11
+ "dtype": "bfloat16",
12
+ "floor_scale": 8192,
13
+ "for_llm_compressor": false,
14
+ "head_dim": 32,
15
+ "hidden_act": "silu",
16
+ "hidden_size": 32,
17
+ "initializer_range": 0.02,
18
+ "interleave_moe_layer_step": 2,
19
+ "intermediate_size": 64,
20
+ "intermediate_size_mlp": 128,
21
+ "layer_types": [
22
+ "chunked_attention",
23
+ "chunked_attention",
24
+ "chunked_attention",
25
+ "full_attention"
26
+ ],
27
+ "max_position_embeddings": 1048576,
28
+ "model_type": "llama4_text",
29
+ "moe_layers": [
30
+ 1,
31
+ 3
32
+ ],
33
+ "neuron": {
34
+ "_serialized_key": "NxDNeuronConfig",
35
+ "batch_size": 1,
36
+ "capacity_factor": null,
37
+ "checkpoint_id": "tiny-random/llama-4",
38
+ "checkpoint_revision": "9e716f5d4d1ffe0a44a15f46f4a12b840439aba4",
39
+ "continuous_batching": false,
40
+ "ep_degree": 1,
41
+ "fused_qkv": false,
42
+ "glu_mlp": true,
43
+ "local_ranks_size": 2,
44
+ "max_batch_size": 1,
45
+ "max_context_length": 1024,
46
+ "max_topk": 256,
47
+ "n_active_tokens": 1024,
48
+ "neuronxcc_version": "2.21.33363.0+82129205",
49
+ "on_device_sampling": true,
50
+ "optimum_neuron_version": "0.4.5.dev1",
51
+ "output_logits": false,
52
+ "pp_degree": 1,
53
+ "sequence_length": 1024,
54
+ "speculation_length": 0,
55
+ "start_rank_id": 0,
56
+ "target": "trn1",
57
+ "torch_dtype": "bfloat16",
58
+ "tp_degree": 2
59
+ },
60
+ "no_rope_layers": [
61
+ 1,
62
+ 1,
63
+ 1,
64
+ 0
65
+ ],
66
+ "num_attention_heads": 1,
67
+ "num_experts_per_tok": 1,
68
+ "num_hidden_layers": 4,
69
+ "num_key_value_heads": 1,
70
+ "num_local_experts": 8,
71
+ "output_router_logits": false,
72
+ "rms_norm_eps": 1e-05,
73
+ "rope_scaling": null,
74
+ "rope_theta": 500000.0,
75
+ "router_aux_loss_coef": 0.001,
76
+ "router_jitter_noise": 0.0,
77
+ "tie_word_embeddings": true,
78
+ "use_cache": true,
79
+ "use_qk_norm": true,
80
+ "vocab_size": 202048
81
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/cf6b9a360dcf294104671106bae2adbd9fd291823bb60a351883163684073231/0aa29066680da68d477e.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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": 1,
22
+ "capacity_factor": null,
23
+ "checkpoint_id": "unsloth/Llama-3.2-1B-Instruct",
24
+ "checkpoint_revision": "5a8abab4a5d6f164389b1079fb721cfab8d7126c",
25
+ "continuous_batching": false,
26
+ "ep_degree": 1,
27
+ "fused_qkv": true,
28
+ "glu_mlp": true,
29
+ "local_ranks_size": 2,
30
+ "max_batch_size": 1,
31
+ "max_context_length": 8192,
32
+ "max_topk": 256,
33
+ "n_active_tokens": 8192,
34
+ "neuronxcc_version": "2.21.33363.0+82129205",
35
+ "on_device_sampling": true,
36
+ "optimum_neuron_version": "0.4.5.dev1",
37
+ "output_logits": false,
38
+ "pp_degree": 1,
39
+ "sequence_length": 8192,
40
+ "speculation_length": 0,
41
+ "start_rank_id": 0,
42
+ "target": "trn1",
43
+ "torch_dtype": "bfloat16",
44
+ "tp_degree": 2
45
+ },
46
+ "num_attention_heads": 32,
47
+ "num_hidden_layers": 16,
48
+ "num_key_value_heads": 8,
49
+ "pretraining_tp": 1,
50
+ "rms_norm_eps": 1e-05,
51
+ "rope_scaling": {
52
+ "factor": 32.0,
53
+ "high_freq_factor": 4.0,
54
+ "low_freq_factor": 1.0,
55
+ "original_max_position_embeddings": 8192,
56
+ "rope_type": "llama3"
57
+ },
58
+ "rope_theta": 500000.0,
59
+ "tie_word_embeddings": true,
60
+ "unsloth_fixed": true,
61
+ "use_cache": true,
62
+ "vocab_size": 128256
63
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/cf6b9a360dcf294104671106bae2adbd9fd291823bb60a351883163684073231/4aab355cf64b87f1cff4.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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": 4,
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": 4,
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": "0.4.5.dev1",
37
+ "output_logits": false,
38
+ "pp_degree": 1,
39
+ "sequence_length": 4096,
40
+ "speculation_length": 0,
41
+ "start_rank_id": 0,
42
+ "target": "trn1",
43
+ "torch_dtype": "bfloat16",
44
+ "tp_degree": 2
45
+ },
46
+ "num_attention_heads": 32,
47
+ "num_hidden_layers": 16,
48
+ "num_key_value_heads": 8,
49
+ "pretraining_tp": 1,
50
+ "rms_norm_eps": 1e-05,
51
+ "rope_scaling": {
52
+ "factor": 32.0,
53
+ "high_freq_factor": 4.0,
54
+ "low_freq_factor": 1.0,
55
+ "original_max_position_embeddings": 8192,
56
+ "rope_type": "llama3"
57
+ },
58
+ "rope_theta": 500000.0,
59
+ "tie_word_embeddings": true,
60
+ "unsloth_fixed": true,
61
+ "use_cache": true,
62
+ "vocab_size": 128256
63
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/cf6b9a360dcf294104671106bae2adbd9fd291823bb60a351883163684073231/72eee6f943925a38948a.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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": 1,
22
+ "capacity_factor": null,
23
+ "checkpoint_id": "unsloth/Llama-3.2-1B-Instruct",
24
+ "checkpoint_revision": null,
25
+ "continuous_batching": false,
26
+ "ep_degree": 1,
27
+ "fused_qkv": false,
28
+ "glu_mlp": true,
29
+ "local_ranks_size": 2,
30
+ "max_batch_size": 1,
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": false,
36
+ "optimum_neuron_version": "0.4.5.dev1",
37
+ "output_logits": false,
38
+ "pp_degree": 1,
39
+ "sequence_length": 4096,
40
+ "speculation_length": 0,
41
+ "start_rank_id": 0,
42
+ "target": "trn1",
43
+ "torch_dtype": "bfloat16",
44
+ "tp_degree": 2
45
+ },
46
+ "num_attention_heads": 32,
47
+ "num_hidden_layers": 16,
48
+ "num_key_value_heads": 8,
49
+ "pretraining_tp": 1,
50
+ "rms_norm_eps": 1e-05,
51
+ "rope_scaling": {
52
+ "factor": 32.0,
53
+ "high_freq_factor": 4.0,
54
+ "low_freq_factor": 1.0,
55
+ "original_max_position_embeddings": 8192,
56
+ "rope_type": "llama3"
57
+ },
58
+ "rope_theta": 500000.0,
59
+ "tie_word_embeddings": true,
60
+ "unsloth_fixed": true,
61
+ "use_cache": true,
62
+ "vocab_size": 128256
63
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/cf6b9a360dcf294104671106bae2adbd9fd291823bb60a351883163684073231/c4a5702b42ed466de9e7.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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": 1,
22
+ "capacity_factor": null,
23
+ "checkpoint_id": "unsloth/Llama-3.2-1B-Instruct",
24
+ "checkpoint_revision": "5a8abab4a5d6f164389b1079fb721cfab8d7126c",
25
+ "continuous_batching": false,
26
+ "ep_degree": 1,
27
+ "fused_qkv": true,
28
+ "glu_mlp": true,
29
+ "local_ranks_size": 2,
30
+ "max_batch_size": 1,
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": "0.4.5.dev1",
37
+ "output_logits": false,
38
+ "pp_degree": 1,
39
+ "sequence_length": 4096,
40
+ "speculation_length": 0,
41
+ "start_rank_id": 0,
42
+ "target": "trn1",
43
+ "torch_dtype": "bfloat16",
44
+ "tp_degree": 2
45
+ },
46
+ "num_attention_heads": 32,
47
+ "num_hidden_layers": 16,
48
+ "num_key_value_heads": 8,
49
+ "pretraining_tp": 1,
50
+ "rms_norm_eps": 1e-05,
51
+ "rope_scaling": {
52
+ "factor": 32.0,
53
+ "high_freq_factor": 4.0,
54
+ "low_freq_factor": 1.0,
55
+ "original_max_position_embeddings": 8192,
56
+ "rope_type": "llama3"
57
+ },
58
+ "rope_theta": 500000.0,
59
+ "tie_word_embeddings": true,
60
+ "unsloth_fixed": true,
61
+ "use_cache": true,
62
+ "vocab_size": 128256
63
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/cf6b9a360dcf294104671106bae2adbd9fd291823bb60a351883163684073231/f18d2067d0df4f48c0c4.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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": 1,
22
+ "capacity_factor": null,
23
+ "checkpoint_id": "unsloth/Llama-3.2-1B-Instruct",
24
+ "checkpoint_revision": null,
25
+ "continuous_batching": false,
26
+ "ep_degree": 1,
27
+ "fused_qkv": false,
28
+ "glu_mlp": true,
29
+ "local_ranks_size": 2,
30
+ "max_batch_size": 1,
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": false,
36
+ "optimum_neuron_version": "0.4.5.dev1",
37
+ "output_logits": false,
38
+ "pp_degree": 1,
39
+ "sequence_length": 4096,
40
+ "speculation_length": 5,
41
+ "start_rank_id": 0,
42
+ "target": "trn1",
43
+ "torch_dtype": "bfloat16",
44
+ "tp_degree": 2
45
+ },
46
+ "num_attention_heads": 32,
47
+ "num_hidden_layers": 16,
48
+ "num_key_value_heads": 8,
49
+ "pretraining_tp": 1,
50
+ "rms_norm_eps": 1e-05,
51
+ "rope_scaling": {
52
+ "factor": 32.0,
53
+ "high_freq_factor": 4.0,
54
+ "low_freq_factor": 1.0,
55
+ "original_max_position_embeddings": 8192,
56
+ "rope_type": "llama3"
57
+ },
58
+ "rope_theta": 500000.0,
59
+ "tie_word_embeddings": true,
60
+ "unsloth_fixed": true,
61
+ "use_cache": true,
62
+ "vocab_size": 128256
63
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/d139acf64685f15794bb983ff6eb881bdd31304bae88b0ce1ed20a54c21f2265/2eeaf9c9a7977232515a.json ADDED
@@ -0,0 +1,58 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "hf-internal-testing/tiny-random-GraniteForCausalLM",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "GraniteForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "attention_multiplier": 1.0,
11
+ "dtype": "float32",
12
+ "embedding_multiplier": 1.0,
13
+ "hidden_act": "silu",
14
+ "hidden_size": 32,
15
+ "initializer_range": 0.02,
16
+ "intermediate_size": 64,
17
+ "logits_scaling": 1.0,
18
+ "max_position_embeddings": 2048,
19
+ "mlp_bias": false,
20
+ "model_type": "granite",
21
+ "neuron": {
22
+ "_serialized_key": "NxDNeuronConfig",
23
+ "batch_size": 1,
24
+ "capacity_factor": null,
25
+ "checkpoint_id": "hf-internal-testing/tiny-random-GraniteForCausalLM",
26
+ "checkpoint_revision": "c3074ebc0ac2fe545305f5e5f6cce2cc9b2aa0c5",
27
+ "continuous_batching": false,
28
+ "ep_degree": 1,
29
+ "fused_qkv": true,
30
+ "glu_mlp": true,
31
+ "local_ranks_size": 2,
32
+ "max_batch_size": 1,
33
+ "max_context_length": 1024,
34
+ "max_topk": 256,
35
+ "n_active_tokens": 1024,
36
+ "neuronxcc_version": "2.21.33363.0+82129205",
37
+ "on_device_sampling": true,
38
+ "optimum_neuron_version": "0.4.5.dev1",
39
+ "output_logits": false,
40
+ "pp_degree": 1,
41
+ "sequence_length": 1024,
42
+ "speculation_length": 0,
43
+ "start_rank_id": 0,
44
+ "target": "trn1",
45
+ "torch_dtype": "float32",
46
+ "tp_degree": 2
47
+ },
48
+ "num_attention_heads": 4,
49
+ "num_hidden_layers": 2,
50
+ "num_key_value_heads": 4,
51
+ "residual_multiplier": 1.0,
52
+ "rms_norm_eps": 1e-06,
53
+ "rope_scaling": null,
54
+ "rope_theta": 10000.0,
55
+ "tie_word_embeddings": false,
56
+ "use_cache": true,
57
+ "vocab_size": 49152
58
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/granite/hf-internal-testing/tiny-random-GraniteForCausalLM/2eeaf9c9a7977232515a.json ADDED
@@ -0,0 +1,58 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "hf-internal-testing/tiny-random-GraniteForCausalLM",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "GraniteForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "attention_multiplier": 1.0,
11
+ "dtype": "float32",
12
+ "embedding_multiplier": 1.0,
13
+ "hidden_act": "silu",
14
+ "hidden_size": 32,
15
+ "initializer_range": 0.02,
16
+ "intermediate_size": 64,
17
+ "logits_scaling": 1.0,
18
+ "max_position_embeddings": 2048,
19
+ "mlp_bias": false,
20
+ "model_type": "granite",
21
+ "neuron": {
22
+ "_serialized_key": "NxDNeuronConfig",
23
+ "batch_size": 1,
24
+ "capacity_factor": null,
25
+ "checkpoint_id": "hf-internal-testing/tiny-random-GraniteForCausalLM",
26
+ "checkpoint_revision": "c3074ebc0ac2fe545305f5e5f6cce2cc9b2aa0c5",
27
+ "continuous_batching": false,
28
+ "ep_degree": 1,
29
+ "fused_qkv": true,
30
+ "glu_mlp": true,
31
+ "local_ranks_size": 2,
32
+ "max_batch_size": 1,
33
+ "max_context_length": 1024,
34
+ "max_topk": 256,
35
+ "n_active_tokens": 1024,
36
+ "neuronxcc_version": "2.21.33363.0+82129205",
37
+ "on_device_sampling": true,
38
+ "optimum_neuron_version": "0.4.5.dev1",
39
+ "output_logits": false,
40
+ "pp_degree": 1,
41
+ "sequence_length": 1024,
42
+ "speculation_length": 0,
43
+ "start_rank_id": 0,
44
+ "target": "trn1",
45
+ "torch_dtype": "float32",
46
+ "tp_degree": 2
47
+ },
48
+ "num_attention_heads": 4,
49
+ "num_hidden_layers": 2,
50
+ "num_key_value_heads": 4,
51
+ "residual_multiplier": 1.0,
52
+ "rms_norm_eps": 1e-06,
53
+ "rope_scaling": null,
54
+ "rope_theta": 10000.0,
55
+ "tie_word_embeddings": false,
56
+ "use_cache": true,
57
+ "vocab_size": 49152
58
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/granite/ibm-granite/granite-3.1-2b-instruct/440dcb82274f114a66ad.json ADDED
@@ -0,0 +1,58 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "ibm-granite/granite-3.1-2b-instruct",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "GraniteForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.1,
10
+ "attention_multiplier": 0.015625,
11
+ "dtype": "bfloat16",
12
+ "embedding_multiplier": 12.0,
13
+ "hidden_act": "silu",
14
+ "hidden_size": 2048,
15
+ "initializer_range": 0.02,
16
+ "intermediate_size": 8192,
17
+ "logits_scaling": 8.0,
18
+ "max_position_embeddings": 131072,
19
+ "mlp_bias": false,
20
+ "model_type": "granite",
21
+ "neuron": {
22
+ "_serialized_key": "NxDNeuronConfig",
23
+ "batch_size": 4,
24
+ "capacity_factor": null,
25
+ "checkpoint_id": "ibm-granite/granite-3.1-2b-instruct",
26
+ "checkpoint_revision": "bbc2aed595bd38bd770263dc3ab831db9794441d",
27
+ "continuous_batching": true,
28
+ "ep_degree": 1,
29
+ "fused_qkv": true,
30
+ "glu_mlp": true,
31
+ "local_ranks_size": 2,
32
+ "max_batch_size": 4,
33
+ "max_context_length": 4096,
34
+ "max_topk": 256,
35
+ "n_active_tokens": 4096,
36
+ "neuronxcc_version": "2.21.33363.0+82129205",
37
+ "on_device_sampling": true,
38
+ "optimum_neuron_version": "0.4.5.dev1",
39
+ "output_logits": false,
40
+ "pp_degree": 1,
41
+ "sequence_length": 4096,
42
+ "speculation_length": 0,
43
+ "start_rank_id": 0,
44
+ "target": "trn1",
45
+ "torch_dtype": "bfloat16",
46
+ "tp_degree": 2
47
+ },
48
+ "num_attention_heads": 32,
49
+ "num_hidden_layers": 40,
50
+ "num_key_value_heads": 8,
51
+ "residual_multiplier": 0.22,
52
+ "rms_norm_eps": 1e-05,
53
+ "rope_scaling": null,
54
+ "rope_theta": 5000000.0,
55
+ "tie_word_embeddings": true,
56
+ "use_cache": true,
57
+ "vocab_size": 49155
58
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/llama/llamafactory/tiny-random-Llama-3/5997ab86579aadcc5cdd.json ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "llamafactory/tiny-random-Llama-3",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "LlamaForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "dtype": "float16",
11
+ "head_dim": 4,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 16,
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 64,
16
+ "max_position_embeddings": 131072,
17
+ "mlp_bias": false,
18
+ "model_type": "llama",
19
+ "neuron": {
20
+ "_serialized_key": "NxDNeuronConfig",
21
+ "batch_size": 1,
22
+ "capacity_factor": null,
23
+ "checkpoint_id": "llamafactory/tiny-random-Llama-3",
24
+ "checkpoint_revision": "bf2a2e3bf199ad2ee96f02a3c00246c608db22a8",
25
+ "continuous_batching": false,
26
+ "ep_degree": 1,
27
+ "fused_qkv": true,
28
+ "glu_mlp": true,
29
+ "local_ranks_size": 2,
30
+ "max_batch_size": 1,
31
+ "max_context_length": 1024,
32
+ "max_topk": 256,
33
+ "n_active_tokens": 1024,
34
+ "neuronxcc_version": "2.21.33363.0+82129205",
35
+ "on_device_sampling": true,
36
+ "optimum_neuron_version": "0.4.5.dev1",
37
+ "output_logits": false,
38
+ "pp_degree": 1,
39
+ "sequence_length": 1024,
40
+ "speculation_length": 0,
41
+ "start_rank_id": 0,
42
+ "target": "trn1",
43
+ "torch_dtype": "float16",
44
+ "tp_degree": 2
45
+ },
46
+ "num_attention_heads": 4,
47
+ "num_hidden_layers": 2,
48
+ "num_key_value_heads": 4,
49
+ "pretraining_tp": 1,
50
+ "rms_norm_eps": 1e-05,
51
+ "rope_scaling": {
52
+ "factor": 8.0,
53
+ "high_freq_factor": 4.0,
54
+ "low_freq_factor": 1.0,
55
+ "original_max_position_embeddings": 8192,
56
+ "rope_type": "llama3"
57
+ },
58
+ "rope_theta": 500000.0,
59
+ "tie_word_embeddings": false,
60
+ "use_cache": true,
61
+ "vocab_size": 128256
62
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/llama/unsloth/Llama-3.2-1B-Instruct/4aab355cf64b87f1cff4.json ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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": 4,
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": 4,
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": "0.4.5.dev1",
37
+ "output_logits": false,
38
+ "pp_degree": 1,
39
+ "sequence_length": 4096,
40
+ "speculation_length": 0,
41
+ "start_rank_id": 0,
42
+ "target": "trn1",
43
+ "torch_dtype": "bfloat16",
44
+ "tp_degree": 2
45
+ },
46
+ "num_attention_heads": 32,
47
+ "num_hidden_layers": 16,
48
+ "num_key_value_heads": 8,
49
+ "pretraining_tp": 1,
50
+ "rms_norm_eps": 1e-05,
51
+ "rope_scaling": {
52
+ "factor": 32.0,
53
+ "high_freq_factor": 4.0,
54
+ "low_freq_factor": 1.0,
55
+ "original_max_position_embeddings": 8192,
56
+ "rope_type": "llama3"
57
+ },
58
+ "rope_theta": 500000.0,
59
+ "tie_word_embeddings": true,
60
+ "unsloth_fixed": true,
61
+ "use_cache": true,
62
+ "vocab_size": 128256
63
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/llama4_text/tiny-random/llama-4/a7d7ccbe80fcc54f090d.json ADDED
@@ -0,0 +1,81 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "tiny-random/llama-4",
4
+ "_task": "text-generation",
5
+ "attention_bias": false,
6
+ "attention_chunk_size": 128,
7
+ "attention_dropout": 0.0,
8
+ "attn_scale": 0.1,
9
+ "attn_temperature_tuning": 4,
10
+ "cache_implementation": "hybrid",
11
+ "dtype": "bfloat16",
12
+ "floor_scale": 8192,
13
+ "for_llm_compressor": false,
14
+ "head_dim": 32,
15
+ "hidden_act": "silu",
16
+ "hidden_size": 32,
17
+ "initializer_range": 0.02,
18
+ "interleave_moe_layer_step": 2,
19
+ "intermediate_size": 64,
20
+ "intermediate_size_mlp": 128,
21
+ "layer_types": [
22
+ "chunked_attention",
23
+ "chunked_attention",
24
+ "chunked_attention",
25
+ "full_attention"
26
+ ],
27
+ "max_position_embeddings": 1048576,
28
+ "model_type": "llama4_text",
29
+ "moe_layers": [
30
+ 1,
31
+ 3
32
+ ],
33
+ "neuron": {
34
+ "_serialized_key": "NxDNeuronConfig",
35
+ "batch_size": 1,
36
+ "capacity_factor": null,
37
+ "checkpoint_id": "tiny-random/llama-4",
38
+ "checkpoint_revision": "9e716f5d4d1ffe0a44a15f46f4a12b840439aba4",
39
+ "continuous_batching": false,
40
+ "ep_degree": 1,
41
+ "fused_qkv": false,
42
+ "glu_mlp": true,
43
+ "local_ranks_size": 2,
44
+ "max_batch_size": 1,
45
+ "max_context_length": 1024,
46
+ "max_topk": 256,
47
+ "n_active_tokens": 1024,
48
+ "neuronxcc_version": "2.21.33363.0+82129205",
49
+ "on_device_sampling": true,
50
+ "optimum_neuron_version": "0.4.5.dev1",
51
+ "output_logits": false,
52
+ "pp_degree": 1,
53
+ "sequence_length": 1024,
54
+ "speculation_length": 0,
55
+ "start_rank_id": 0,
56
+ "target": "trn1",
57
+ "torch_dtype": "bfloat16",
58
+ "tp_degree": 2
59
+ },
60
+ "no_rope_layers": [
61
+ 1,
62
+ 1,
63
+ 1,
64
+ 0
65
+ ],
66
+ "num_attention_heads": 1,
67
+ "num_experts_per_tok": 1,
68
+ "num_hidden_layers": 4,
69
+ "num_key_value_heads": 1,
70
+ "num_local_experts": 8,
71
+ "output_router_logits": false,
72
+ "rms_norm_eps": 1e-05,
73
+ "rope_scaling": null,
74
+ "rope_theta": 500000.0,
75
+ "router_aux_loss_coef": 0.001,
76
+ "router_jitter_noise": 0.0,
77
+ "tie_word_embeddings": true,
78
+ "use_cache": true,
79
+ "use_qk_norm": true,
80
+ "vocab_size": 202048
81
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/mixtral/dacorvo/Mixtral-tiny/93ac0e7f7bd685a9d58b.json ADDED
@@ -0,0 +1,58 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "dacorvo/Mixtral-tiny",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "MixtralForCausalLM"
7
+ ],
8
+ "attention_dropout": 0.0,
9
+ "dtype": "float16",
10
+ "head_dim": 32,
11
+ "hidden_act": "silu",
12
+ "hidden_size": 1024,
13
+ "initializer_range": 0.02,
14
+ "intermediate_size": 3584,
15
+ "max_position_embeddings": 1024,
16
+ "model_type": "mixtral",
17
+ "neuron": {
18
+ "_serialized_key": "NxDNeuronConfig",
19
+ "batch_size": 1,
20
+ "capacity_factor": null,
21
+ "checkpoint_id": "dacorvo/Mixtral-tiny",
22
+ "checkpoint_revision": "c557ba205ddff6ea911f4719e0d543d6c08356b6",
23
+ "continuous_batching": false,
24
+ "ep_degree": 1,
25
+ "fused_qkv": false,
26
+ "glu_mlp": true,
27
+ "local_ranks_size": 2,
28
+ "max_batch_size": 1,
29
+ "max_context_length": 1024,
30
+ "max_topk": 256,
31
+ "n_active_tokens": 1024,
32
+ "neuronxcc_version": "2.21.33363.0+82129205",
33
+ "on_device_sampling": false,
34
+ "optimum_neuron_version": "0.4.5.dev1",
35
+ "output_logits": false,
36
+ "pp_degree": 1,
37
+ "sequence_length": 1024,
38
+ "speculation_length": 0,
39
+ "start_rank_id": 0,
40
+ "target": "trn1",
41
+ "torch_dtype": "float16",
42
+ "tp_degree": 2
43
+ },
44
+ "num_attention_heads": 32,
45
+ "num_experts_per_tok": 2,
46
+ "num_hidden_layers": 2,
47
+ "num_key_value_heads": 8,
48
+ "num_local_experts": 8,
49
+ "output_router_logits": false,
50
+ "rms_norm_eps": 1e-05,
51
+ "rope_theta": 10000.0,
52
+ "router_aux_loss_coef": 0.001,
53
+ "router_jitter_noise": 0.0,
54
+ "sliding_window": 4096,
55
+ "tie_word_embeddings": false,
56
+ "use_cache": true,
57
+ "vocab_size": 32000
58
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/phi3/microsoft/Phi-3.5-mini-instruct/8cefa5ff9083418aa040.json ADDED
@@ -0,0 +1,164 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "microsoft/Phi-3.5-mini-instruct",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "Phi3ForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "auto_map": {
11
+ "AutoConfig": "configuration_phi3.Phi3Config",
12
+ "AutoModelForCausalLM": "modeling_phi3.Phi3ForCausalLM"
13
+ },
14
+ "dtype": "bfloat16",
15
+ "embd_pdrop": 0.0,
16
+ "hidden_act": "silu",
17
+ "hidden_size": 3072,
18
+ "initializer_range": 0.02,
19
+ "intermediate_size": 8192,
20
+ "max_position_embeddings": 131072,
21
+ "model_type": "phi3",
22
+ "neuron": {
23
+ "_serialized_key": "NxDNeuronConfig",
24
+ "batch_size": 4,
25
+ "capacity_factor": null,
26
+ "checkpoint_id": "microsoft/Phi-3.5-mini-instruct",
27
+ "checkpoint_revision": "2fe192450127e6a83f7441aef6e3ca586c338b77",
28
+ "continuous_batching": true,
29
+ "ep_degree": 1,
30
+ "fused_qkv": true,
31
+ "glu_mlp": true,
32
+ "local_ranks_size": 2,
33
+ "max_batch_size": 4,
34
+ "max_context_length": 4096,
35
+ "max_topk": 256,
36
+ "n_active_tokens": 4096,
37
+ "neuronxcc_version": "2.21.33363.0+82129205",
38
+ "on_device_sampling": true,
39
+ "optimum_neuron_version": "0.4.5.dev1",
40
+ "output_logits": false,
41
+ "pp_degree": 1,
42
+ "sequence_length": 4096,
43
+ "speculation_length": 0,
44
+ "start_rank_id": 0,
45
+ "target": "trn1",
46
+ "torch_dtype": "bfloat16",
47
+ "tp_degree": 2
48
+ },
49
+ "num_attention_heads": 32,
50
+ "num_hidden_layers": 32,
51
+ "num_key_value_heads": 32,
52
+ "original_max_position_embeddings": 4096,
53
+ "partial_rotary_factor": 1.0,
54
+ "resid_pdrop": 0.0,
55
+ "rms_norm_eps": 1e-05,
56
+ "rope_scaling": {
57
+ "long_factor": [
58
+ 1.0800000429153442,
59
+ 1.1100000143051147,
60
+ 1.1399999856948853,
61
+ 1.340000033378601,
62
+ 1.5899999141693115,
63
+ 1.600000023841858,
64
+ 1.6200000047683716,
65
+ 2.620000123977661,
66
+ 3.2300000190734863,
67
+ 3.2300000190734863,
68
+ 4.789999961853027,
69
+ 7.400000095367432,
70
+ 7.700000286102295,
71
+ 9.09000015258789,
72
+ 12.199999809265137,
73
+ 17.670000076293945,
74
+ 24.46000099182129,
75
+ 28.57000160217285,
76
+ 30.420001983642578,
77
+ 30.840002059936523,
78
+ 32.590003967285156,
79
+ 32.93000411987305,
80
+ 42.320003509521484,
81
+ 44.96000289916992,
82
+ 50.340003967285156,
83
+ 50.45000457763672,
84
+ 57.55000305175781,
85
+ 57.93000411987305,
86
+ 58.21000289916992,
87
+ 60.1400032043457,
88
+ 62.61000442504883,
89
+ 62.62000274658203,
90
+ 62.71000289916992,
91
+ 63.1400032043457,
92
+ 63.1400032043457,
93
+ 63.77000427246094,
94
+ 63.93000411987305,
95
+ 63.96000289916992,
96
+ 63.970001220703125,
97
+ 64.02999877929688,
98
+ 64.06999969482422,
99
+ 64.08000183105469,
100
+ 64.12000274658203,
101
+ 64.41000366210938,
102
+ 64.4800033569336,
103
+ 64.51000213623047,
104
+ 64.52999877929688,
105
+ 64.83999633789062
106
+ ],
107
+ "short_factor": [
108
+ 1.0,
109
+ 1.0199999809265137,
110
+ 1.0299999713897705,
111
+ 1.0299999713897705,
112
+ 1.0499999523162842,
113
+ 1.0499999523162842,
114
+ 1.0499999523162842,
115
+ 1.0499999523162842,
116
+ 1.0499999523162842,
117
+ 1.0699999332427979,
118
+ 1.0999999046325684,
119
+ 1.1099998950958252,
120
+ 1.1599998474121094,
121
+ 1.1599998474121094,
122
+ 1.1699998378753662,
123
+ 1.2899998426437378,
124
+ 1.339999794960022,
125
+ 1.679999828338623,
126
+ 1.7899998426437378,
127
+ 1.8199998140335083,
128
+ 1.8499997854232788,
129
+ 1.8799997568130493,
130
+ 1.9099997282028198,
131
+ 1.9399996995925903,
132
+ 1.9899996519088745,
133
+ 2.0199997425079346,
134
+ 2.0199997425079346,
135
+ 2.0199997425079346,
136
+ 2.0199997425079346,
137
+ 2.0199997425079346,
138
+ 2.0199997425079346,
139
+ 2.0299997329711914,
140
+ 2.0299997329711914,
141
+ 2.0299997329711914,
142
+ 2.0299997329711914,
143
+ 2.0299997329711914,
144
+ 2.0299997329711914,
145
+ 2.0299997329711914,
146
+ 2.0299997329711914,
147
+ 2.0299997329711914,
148
+ 2.0799996852874756,
149
+ 2.0899996757507324,
150
+ 2.189999580383301,
151
+ 2.2199995517730713,
152
+ 2.5899994373321533,
153
+ 2.729999542236328,
154
+ 2.749999523162842,
155
+ 2.8399994373321533
156
+ ],
157
+ "type": "longrope"
158
+ },
159
+ "rope_theta": 10000.0,
160
+ "sliding_window": 262144,
161
+ "tie_word_embeddings": false,
162
+ "use_cache": true,
163
+ "vocab_size": 32064
164
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/phi3/yujiepan/phi-4-tiny-random/4925f3a03aeddc47881d.json ADDED
@@ -0,0 +1,59 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "yujiepan/phi-4-tiny-random",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "Phi3ForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "auto_map": {},
11
+ "dtype": "bfloat16",
12
+ "embd_pdrop": 0.0,
13
+ "hidden_act": "silu",
14
+ "hidden_size": 16,
15
+ "initializer_range": 0.02,
16
+ "intermediate_size": 32,
17
+ "max_position_embeddings": 16384,
18
+ "model_type": "phi3",
19
+ "neuron": {
20
+ "_serialized_key": "NxDNeuronConfig",
21
+ "batch_size": 1,
22
+ "capacity_factor": null,
23
+ "checkpoint_id": "yujiepan/phi-4-tiny-random",
24
+ "checkpoint_revision": "18a9a1168dc97ac6d128f811925670c275610f5a",
25
+ "continuous_batching": false,
26
+ "ep_degree": 1,
27
+ "fused_qkv": true,
28
+ "glu_mlp": true,
29
+ "local_ranks_size": 2,
30
+ "max_batch_size": 1,
31
+ "max_context_length": 1024,
32
+ "max_topk": 256,
33
+ "n_active_tokens": 1024,
34
+ "neuronxcc_version": "2.21.33363.0+82129205",
35
+ "on_device_sampling": true,
36
+ "optimum_neuron_version": "0.4.5.dev1",
37
+ "output_logits": false,
38
+ "pp_degree": 1,
39
+ "sequence_length": 1024,
40
+ "speculation_length": 0,
41
+ "start_rank_id": 0,
42
+ "target": "trn1",
43
+ "torch_dtype": "bfloat16",
44
+ "tp_degree": 2
45
+ },
46
+ "num_attention_heads": 2,
47
+ "num_hidden_layers": 2,
48
+ "num_key_value_heads": 1,
49
+ "original_max_position_embeddings": 16384,
50
+ "partial_rotary_factor": 1.0,
51
+ "resid_pdrop": 0.0,
52
+ "rms_norm_eps": 1e-05,
53
+ "rope_scaling": null,
54
+ "rope_theta": 250000,
55
+ "sliding_window": null,
56
+ "tie_word_embeddings": false,
57
+ "use_cache": true,
58
+ "vocab_size": 100352
59
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/qwen2/Qwen/Qwen2.5-0.5B/5ec0bcb03ab4e2ad844c.json ADDED
@@ -0,0 +1,82 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "Qwen/Qwen2.5-0.5B",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "Qwen2ForCausalLM"
7
+ ],
8
+ "attention_dropout": 0.0,
9
+ "dtype": "bfloat16",
10
+ "hidden_act": "silu",
11
+ "hidden_size": 896,
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 4864,
14
+ "layer_types": [
15
+ "full_attention",
16
+ "full_attention",
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
+ ],
40
+ "max_position_embeddings": 32768,
41
+ "max_window_layers": 24,
42
+ "model_type": "qwen2",
43
+ "neuron": {
44
+ "_serialized_key": "NxDNeuronConfig",
45
+ "batch_size": 4,
46
+ "capacity_factor": null,
47
+ "checkpoint_id": "Qwen/Qwen2.5-0.5B",
48
+ "checkpoint_revision": "060db6499f32faf8b98477b0a26969ef7d8b9987",
49
+ "continuous_batching": true,
50
+ "ep_degree": 1,
51
+ "fused_qkv": false,
52
+ "glu_mlp": true,
53
+ "local_ranks_size": 2,
54
+ "max_batch_size": 4,
55
+ "max_context_length": 4096,
56
+ "max_topk": 256,
57
+ "n_active_tokens": 4096,
58
+ "neuronxcc_version": "2.21.33363.0+82129205",
59
+ "on_device_sampling": false,
60
+ "optimum_neuron_version": "0.4.5.dev1",
61
+ "output_logits": false,
62
+ "pp_degree": 1,
63
+ "sequence_length": 4096,
64
+ "speculation_length": 0,
65
+ "start_rank_id": 0,
66
+ "target": "trn1",
67
+ "torch_dtype": "bfloat16",
68
+ "tp_degree": 2
69
+ },
70
+ "num_attention_heads": 14,
71
+ "num_hidden_layers": 24,
72
+ "num_key_value_heads": 2,
73
+ "rms_norm_eps": 1e-06,
74
+ "rope_scaling": null,
75
+ "rope_theta": 1000000.0,
76
+ "sliding_window": null,
77
+ "tie_word_embeddings": true,
78
+ "use_cache": true,
79
+ "use_mrope": false,
80
+ "use_sliding_window": false,
81
+ "vocab_size": 151936
82
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/qwen2/yujiepan/qwen2.5-128k-tiny-random/a037d3d5582033b493b7.json ADDED
@@ -0,0 +1,64 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "yujiepan/qwen2.5-128k-tiny-random",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "Qwen2ForCausalLM"
7
+ ],
8
+ "attention_dropout": 0.0,
9
+ "dtype": "bfloat16",
10
+ "hidden_act": "silu",
11
+ "hidden_size": 8,
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 16,
14
+ "layer_types": [
15
+ "full_attention",
16
+ "full_attention"
17
+ ],
18
+ "max_position_embeddings": 32768,
19
+ "max_window_layers": 1,
20
+ "model_type": "qwen2",
21
+ "neuron": {
22
+ "_serialized_key": "NxDNeuronConfig",
23
+ "batch_size": 1,
24
+ "capacity_factor": null,
25
+ "checkpoint_id": "yujiepan/qwen2.5-128k-tiny-random",
26
+ "checkpoint_revision": "c8296d4ca3f87782876d2382fbb6481d1beb8ef0",
27
+ "continuous_batching": false,
28
+ "ep_degree": 1,
29
+ "fused_qkv": false,
30
+ "glu_mlp": true,
31
+ "local_ranks_size": 2,
32
+ "max_batch_size": 1,
33
+ "max_context_length": 1024,
34
+ "max_topk": 256,
35
+ "n_active_tokens": 1024,
36
+ "neuronxcc_version": "2.21.33363.0+82129205",
37
+ "on_device_sampling": true,
38
+ "optimum_neuron_version": "0.4.5.dev1",
39
+ "output_logits": false,
40
+ "pp_degree": 1,
41
+ "sequence_length": 1024,
42
+ "speculation_length": 0,
43
+ "start_rank_id": 0,
44
+ "target": "trn1",
45
+ "torch_dtype": "bfloat16",
46
+ "tp_degree": 2
47
+ },
48
+ "num_attention_heads": 4,
49
+ "num_hidden_layers": 2,
50
+ "num_key_value_heads": 2,
51
+ "rms_norm_eps": 1e-06,
52
+ "rope_scaling": {
53
+ "factor": 4.0,
54
+ "original_max_position_embeddings": 32768,
55
+ "rope_type": "yarn",
56
+ "type": "yarn"
57
+ },
58
+ "rope_theta": 1000000.0,
59
+ "sliding_window": null,
60
+ "tie_word_embeddings": false,
61
+ "use_cache": true,
62
+ "use_sliding_window": false,
63
+ "vocab_size": 152064
64
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/qwen3/Qwen/Qwen3-Embedding-0.6B/2c2bad80f80ab55a2d84.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "Qwen/Qwen3-Embedding-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": 32768,
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-Embedding-0.6B",
54
+ "checkpoint_revision": "c54f2e6e80b2d7b7de06f51cec4959f6b3e03418",
55
+ "continuous_batching": false,
56
+ "ep_degree": 1,
57
+ "fused_qkv": true,
58
+ "glu_mlp": true,
59
+ "local_ranks_size": 2,
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": false,
66
+ "optimum_neuron_version": "0.4.5.dev1",
67
+ "output_logits": false,
68
+ "pp_degree": 1,
69
+ "sequence_length": 1024,
70
+ "speculation_length": 0,
71
+ "start_rank_id": 0,
72
+ "target": "trn1",
73
+ "torch_dtype": "bfloat16",
74
+ "tp_degree": 2
75
+ },
76
+ "num_attention_heads": 16,
77
+ "num_hidden_layers": 28,
78
+ "num_key_value_heads": 8,
79
+ "rms_norm_eps": 1e-06,
80
+ "rope_scaling": null,
81
+ "rope_theta": 1000000,
82
+ "sliding_window": null,
83
+ "tie_word_embeddings": true,
84
+ "use_cache": true,
85
+ "use_sliding_window": false,
86
+ "vocab_size": 151669
87
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/qwen3_moe/optimum-internal-testing/tiny-random-qwen3_moe/ada37737b07aadde24a9.json ADDED
@@ -0,0 +1,65 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "optimum-internal-testing/tiny-random-qwen3_moe",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "Qwen3MoeForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "decoder_sparse_step": 2,
11
+ "dtype": "float32",
12
+ "head_dim": 32,
13
+ "hidden_act": "silu",
14
+ "hidden_size": 64,
15
+ "initializer_range": 0.02,
16
+ "intermediate_size": 128,
17
+ "max_position_embeddings": 40960,
18
+ "max_window_layers": 1,
19
+ "mlp_only_layers": [],
20
+ "model_type": "qwen3_moe",
21
+ "moe_intermediate_size": 128,
22
+ "neuron": {
23
+ "_serialized_key": "NxDNeuronConfig",
24
+ "batch_size": 1,
25
+ "capacity_factor": null,
26
+ "checkpoint_id": "optimum-internal-testing/tiny-random-qwen3_moe",
27
+ "checkpoint_revision": "e0230be2839556b44b7400a233c73c74b4abb7af",
28
+ "continuous_batching": false,
29
+ "ep_degree": 1,
30
+ "fused_qkv": false,
31
+ "glu_mlp": true,
32
+ "local_ranks_size": 2,
33
+ "max_batch_size": 1,
34
+ "max_context_length": 1024,
35
+ "max_topk": 256,
36
+ "n_active_tokens": 1024,
37
+ "neuronxcc_version": "2.21.33363.0+82129205",
38
+ "on_device_sampling": true,
39
+ "optimum_neuron_version": "0.4.5.dev1",
40
+ "output_logits": false,
41
+ "pp_degree": 1,
42
+ "sequence_length": 1024,
43
+ "speculation_length": 0,
44
+ "start_rank_id": 0,
45
+ "target": "trn1",
46
+ "torch_dtype": "float32",
47
+ "tp_degree": 2
48
+ },
49
+ "norm_topk_prob": true,
50
+ "num_attention_heads": 2,
51
+ "num_experts": 8,
52
+ "num_experts_per_tok": 2,
53
+ "num_hidden_layers": 2,
54
+ "num_key_value_heads": 1,
55
+ "output_router_logits": false,
56
+ "rms_norm_eps": 1e-06,
57
+ "rope_scaling": null,
58
+ "rope_theta": 1000000.0,
59
+ "router_aux_loss_coef": 0.001,
60
+ "sliding_window": null,
61
+ "tie_word_embeddings": true,
62
+ "use_cache": true,
63
+ "use_sliding_window": false,
64
+ "vocab_size": 151936
65
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev1/smollm3/HuggingFaceTB/SmolLM3-3B/7144c43faca4fea18b01.json ADDED
@@ -0,0 +1,134 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "HuggingFaceTB/SmolLM3-3B",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "SmolLM3ForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "dtype": "bfloat16",
11
+ "hidden_act": "silu",
12
+ "hidden_size": 2048,
13
+ "initializer_range": 0.02,
14
+ "intermediate_size": 11008,
15
+ "layer_types": [
16
+ "full_attention",
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
+ ],
53
+ "max_position_embeddings": 65536,
54
+ "max_window_layers": 28,
55
+ "mlp_bias": false,
56
+ "model_type": "smollm3",
57
+ "neuron": {
58
+ "_serialized_key": "NxDNeuronConfig",
59
+ "batch_size": 4,
60
+ "capacity_factor": null,
61
+ "checkpoint_id": "HuggingFaceTB/SmolLM3-3B",
62
+ "checkpoint_revision": "a07cc9a04f16550a088caea529712d1d335b0ac1",
63
+ "continuous_batching": true,
64
+ "ep_degree": 1,
65
+ "fused_qkv": true,
66
+ "glu_mlp": true,
67
+ "local_ranks_size": 2,
68
+ "max_batch_size": 4,
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": true,
74
+ "optimum_neuron_version": "0.4.5.dev1",
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
+ "no_rope_layer_interval": 4,
85
+ "no_rope_layers": [
86
+ 1,
87
+ 1,
88
+ 1,
89
+ 0,
90
+ 1,
91
+ 1,
92
+ 1,
93
+ 0,
94
+ 1,
95
+ 1,
96
+ 1,
97
+ 0,
98
+ 1,
99
+ 1,
100
+ 1,
101
+ 0,
102
+ 1,
103
+ 1,
104
+ 1,
105
+ 0,
106
+ 1,
107
+ 1,
108
+ 1,
109
+ 0,
110
+ 1,
111
+ 1,
112
+ 1,
113
+ 0,
114
+ 1,
115
+ 1,
116
+ 1,
117
+ 0,
118
+ 1,
119
+ 1,
120
+ 1,
121
+ 0
122
+ ],
123
+ "num_attention_heads": 16,
124
+ "num_hidden_layers": 36,
125
+ "num_key_value_heads": 4,
126
+ "pretraining_tp": 2,
127
+ "rms_norm_eps": 1e-06,
128
+ "rope_scaling": null,
129
+ "rope_theta": 5000000.0,
130
+ "sliding_window": null,
131
+ "use_cache": false,
132
+ "use_sliding_window": false,
133
+ "vocab_size": 128256
134
+ }
neuronxcc-2.21.33363.0+82129205/MODULE_04d46d63afe5b4e1584b+6bc3ae29/compile_flags.json ADDED
@@ -0,0 +1 @@
 
 
1
+ ["--target=trn1", "--enable-saturate-infinity", "--auto-cast=none", "--model-type=transformer", "-O1", "--logfile=/tmp/nxdi_test_f86db28f-8362-4ab3-bca1-87fb7357b2a1/compiler_workdir/AttentionModelWrapper/_tp0_bk0/log-neuron-cc.txt", "--enable-internal-neff-wrapper"]
neuronxcc-2.21.33363.0+82129205/MODULE_04d46d63afe5b4e1584b+6bc3ae29/model.done ADDED
File without changes
neuronxcc-2.21.33363.0+82129205/MODULE_04d46d63afe5b4e1584b+6bc3ae29/model.hlo_module.pb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:be2fc82d01eae877238b23e1fa62d9bd75d648e5ffdf58b8cf01e4f672c3e27e
3
+ size 11280
neuronxcc-2.21.33363.0+82129205/MODULE_04d46d63afe5b4e1584b+6bc3ae29/model.neff ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:03bae6db1a714232a6d34ce6ceeec0ed6626740f78e3af4b5003cc96596430c5
3
+ size 1444864
neuronxcc-2.21.33363.0+82129205/MODULE_04d46d63afe5b4e1584b+6bc3ae29/wrapped_neff.hlo ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:afe8ea4bd2198e83af2c041364b064a638c2aea8ab17aa34c467e4bb7c5b4724
3
+ size 1447845
neuronxcc-2.21.33363.0+82129205/MODULE_10217061096959125489+e30acd3a/model.neff CHANGED
Binary files a/neuronxcc-2.21.33363.0+82129205/MODULE_10217061096959125489+e30acd3a/model.neff and b/neuronxcc-2.21.33363.0+82129205/MODULE_10217061096959125489+e30acd3a/model.neff differ
 
neuronxcc-2.21.33363.0+82129205/MODULE_10244305442015770634+e30acd3a/model.neff CHANGED
Binary files a/neuronxcc-2.21.33363.0+82129205/MODULE_10244305442015770634+e30acd3a/model.neff and b/neuronxcc-2.21.33363.0+82129205/MODULE_10244305442015770634+e30acd3a/model.neff differ
 
neuronxcc-2.21.33363.0+82129205/MODULE_10606948783918825529+e30acd3a/model.neff CHANGED
Binary files a/neuronxcc-2.21.33363.0+82129205/MODULE_10606948783918825529+e30acd3a/model.neff and b/neuronxcc-2.21.33363.0+82129205/MODULE_10606948783918825529+e30acd3a/model.neff differ
 
neuronxcc-2.21.33363.0+82129205/MODULE_10645643398657092095+e30acd3a/model.neff CHANGED
Binary files a/neuronxcc-2.21.33363.0+82129205/MODULE_10645643398657092095+e30acd3a/model.neff and b/neuronxcc-2.21.33363.0+82129205/MODULE_10645643398657092095+e30acd3a/model.neff differ