Hugging Face's logo

aws-neuron
/
optimum-neuron-cache

Model card Files
xet
 Community
665
tengomucho HF Staff commited on
Commit
09eda83
·
verified ·
1 Parent(s): e597ee3

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 +49 -0
  2. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/0cc25526e2cfc37a8875a3752f33c4d7505d8a07b869d0f3f41915cf6e763b74/8a4430107c8609f149e2.json +58 -0
  3. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/0cc25526e2cfc37a8875a3752f33c4d7505d8a07b869d0f3f41915cf6e763b74/de33f9f75b62f508e84e.json +58 -0
  4. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/441269935591cad8d370e512c0b93cdd2fce6247c40e5a4866d872ee5338b0de/8d757b039a90a33816e7.json +81 -0
  5. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/4cb7aff9e2a15c151396f2b684013e39d6739f0dec83e5c9dabbfe9d5fcf77b7/287ad99b2a6141ebb12e.json +82 -0
  6. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/4cb7aff9e2a15c151396f2b684013e39d6739f0dec83e5c9dabbfe9d5fcf77b7/7911379237ca2e32dc36.json +82 -0
  7. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/60feecaa0c4c075e2f3e46a3f55d9a273f0ddd75a0ecf64e4ae27352e0819506/0b9d5658f10ecb783ea7.json +87 -0
  8. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/60feecaa0c4c075e2f3e46a3f55d9a273f0ddd75a0ecf64e4ae27352e0819506/3aace2df87913058a524.json +87 -0
  9. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/73707b485eab9008c7aba7f5dad0ce2384ac685318d5f888c12fa0d81ed90b19/8ff57fd8c6031855b0ec.json +134 -0
  10. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/73707b485eab9008c7aba7f5dad0ce2384ac685318d5f888c12fa0d81ed90b19/c100f806a6006feef39b.json +134 -0
  11. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/8c90ac2593ed0b7f1ecb60e82cb184fb11f2ea640befa1cc7b10766a5c02525d/3943e7730d04a5ca4a9d.json +164 -0
  12. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/8c90ac2593ed0b7f1ecb60e82cb184fb11f2ea640befa1cc7b10766a5c02525d/fa7402288853e31635c0.json +164 -0
  13. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/929b02754a13cbfdf657d863c3fc6f3bce672879bc6ae48ab45be21e881e9ec2/5b6c82d29f13b10e5052.json +87 -0
  14. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/929b02754a13cbfdf657d863c3fc6f3bce672879bc6ae48ab45be21e881e9ec2/c4069e9e5ecd272fe629.json +87 -0
  15. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/bd8b5aaeb4693c199f97f6eebd1bbca77255bea6e0ed042c0c592ee5adccec97/262dbf43da8b9b831afa.json +80 -0
  16. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/bd8b5aaeb4693c199f97f6eebd1bbca77255bea6e0ed042c0c592ee5adccec97/a5218019da46cb49d020.json +80 -0
  17. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/bd8b5aaeb4693c199f97f6eebd1bbca77255bea6e0ed042c0c592ee5adccec97/c5893ac2150961103e95.json +80 -0
  18. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/cf6b9a360dcf294104671106bae2adbd9fd291823bb60a351883163684073231/02e21fa4187e1718bddb.json +63 -0
  19. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/cf6b9a360dcf294104671106bae2adbd9fd291823bb60a351883163684073231/b348680583357cdc479f.json +63 -0
  20. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/gemma3_text/google/gemma-3-270m-it/c5893ac2150961103e95.json +80 -0
  21. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/gemma3_text/unsloth/gemma-3-270m-it/8d757b039a90a33816e7.json +81 -0
  22. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/granite/ibm-granite/granite-3.1-2b-instruct/8a4430107c8609f149e2.json +58 -0
  23. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/llama/unsloth/Llama-3.2-1B-Instruct/b348680583357cdc479f.json +63 -0
  24. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/phi3/microsoft/Phi-3.5-mini-instruct/3943e7730d04a5ca4a9d.json +164 -0
  25. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/qwen2/Qwen/Qwen2.5-0.5B/287ad99b2a6141ebb12e.json +82 -0
  26. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/qwen3/Qwen/Qwen3-0.6B/c4069e9e5ecd272fe629.json +87 -0
  27. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/qwen3/Qwen/Qwen3-Embedding-0.6B/0b9d5658f10ecb783ea7.json +87 -0
  28. neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/smollm3/HuggingFaceTB/SmolLM3-3B/c100f806a6006feef39b.json +134 -0
  29. neuronxcc-2.21.33363.0+82129205/MODULE_0e8dfbbde719dcf9a38a+a32116a7/compile_flags.json +1 -0
  30. neuronxcc-2.21.33363.0+82129205/MODULE_0e8dfbbde719dcf9a38a+a32116a7/model.done +0 -0
  31. neuronxcc-2.21.33363.0+82129205/MODULE_0e8dfbbde719dcf9a38a+a32116a7/model.hlo_module.pb +3 -0
  32. neuronxcc-2.21.33363.0+82129205/MODULE_0e8dfbbde719dcf9a38a+a32116a7/model.neff +3 -0
  33. neuronxcc-2.21.33363.0+82129205/MODULE_164f48a2a338a277ba9b+a32116a7/compile_flags.json +1 -0
  34. neuronxcc-2.21.33363.0+82129205/MODULE_164f48a2a338a277ba9b+a32116a7/model.done +0 -0
  35. neuronxcc-2.21.33363.0+82129205/MODULE_164f48a2a338a277ba9b+a32116a7/model.hlo_module.pb +3 -0
  36. neuronxcc-2.21.33363.0+82129205/MODULE_164f48a2a338a277ba9b+a32116a7/model.neff +3 -0
  37. neuronxcc-2.21.33363.0+82129205/MODULE_169cfc2b665743de4e2b+ac10809c/compile_flags.json +1 -0
  38. neuronxcc-2.21.33363.0+82129205/MODULE_169cfc2b665743de4e2b+ac10809c/model.done +0 -0
  39. neuronxcc-2.21.33363.0+82129205/MODULE_169cfc2b665743de4e2b+ac10809c/model.hlo_module.pb +3 -0
  40. neuronxcc-2.21.33363.0+82129205/MODULE_169cfc2b665743de4e2b+ac10809c/model.neff +3 -0
  41. neuronxcc-2.21.33363.0+82129205/MODULE_169cfc2b665743de4e2b+ac10809c/wrapped_neff.hlo +3 -0
  42. neuronxcc-2.21.33363.0+82129205/MODULE_18ad931622c6ffd3a96c+ac10809c/compile_flags.json +1 -0
  43. neuronxcc-2.21.33363.0+82129205/MODULE_18ad931622c6ffd3a96c+ac10809c/model.done +0 -0
  44. neuronxcc-2.21.33363.0+82129205/MODULE_18ad931622c6ffd3a96c+ac10809c/model.hlo_module.pb +3 -0
  45. neuronxcc-2.21.33363.0+82129205/MODULE_18ad931622c6ffd3a96c+ac10809c/model.neff +3 -0
  46. neuronxcc-2.21.33363.0+82129205/MODULE_18ad931622c6ffd3a96c+ac10809c/wrapped_neff.hlo +3 -0
  47. neuronxcc-2.21.33363.0+82129205/MODULE_1d5235b31814f7b0b086+a32116a7/compile_flags.json +1 -0
  48. neuronxcc-2.21.33363.0+82129205/MODULE_1d5235b31814f7b0b086+a32116a7/model.done +0 -0
  49. neuronxcc-2.21.33363.0+82129205/MODULE_1d5235b31814f7b0b086+a32116a7/model.hlo_module.pb +3 -0
  50. neuronxcc-2.21.33363.0+82129205/MODULE_1d5235b31814f7b0b086+a32116a7/model.neff +3 -0
.gitattributes CHANGED
@@ -16503,3 +16503,52 @@ neuronxcc-2.21.33363.0+82129205/MODULE_fa44a5acf289ee1ecd41+8ee4e2e4/model.neff
16503
  neuronxcc-2.21.33363.0+82129205/MODULE_fa44a5acf289ee1ecd41+8ee4e2e4/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
16504
  neuronxcc-2.21.33363.0+82129205/MODULE_e7fddd20b107d5347811+fb4cc044/model.neff filter=lfs diff=lfs merge=lfs -text
16505
  neuronxcc-2.21.33363.0+82129205/MODULE_84fd59187fd6be117f3a+fb4cc044/model.neff filter=lfs diff=lfs merge=lfs -text
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
16503
  neuronxcc-2.21.33363.0+82129205/MODULE_fa44a5acf289ee1ecd41+8ee4e2e4/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
16504
  neuronxcc-2.21.33363.0+82129205/MODULE_e7fddd20b107d5347811+fb4cc044/model.neff filter=lfs diff=lfs merge=lfs -text
16505
  neuronxcc-2.21.33363.0+82129205/MODULE_84fd59187fd6be117f3a+fb4cc044/model.neff filter=lfs diff=lfs merge=lfs -text
16506
+ neuronxcc-2.21.33363.0+82129205/MODULE_0e8dfbbde719dcf9a38a+a32116a7/model.neff filter=lfs diff=lfs merge=lfs -text
16507
+ neuronxcc-2.21.33363.0+82129205/MODULE_164f48a2a338a277ba9b+a32116a7/model.neff filter=lfs diff=lfs merge=lfs -text
16508
+ neuronxcc-2.21.33363.0+82129205/MODULE_169cfc2b665743de4e2b+ac10809c/model.neff filter=lfs diff=lfs merge=lfs -text
16509
+ neuronxcc-2.21.33363.0+82129205/MODULE_169cfc2b665743de4e2b+ac10809c/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
16510
+ neuronxcc-2.21.33363.0+82129205/MODULE_18ad931622c6ffd3a96c+ac10809c/model.neff filter=lfs diff=lfs merge=lfs -text
16511
+ neuronxcc-2.21.33363.0+82129205/MODULE_18ad931622c6ffd3a96c+ac10809c/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
16512
+ neuronxcc-2.21.33363.0+82129205/MODULE_1d5235b31814f7b0b086+a32116a7/model.neff filter=lfs diff=lfs merge=lfs -text
16513
+ neuronxcc-2.21.33363.0+82129205/MODULE_248be7993b4edc8d46c8+ac10809c/model.neff filter=lfs diff=lfs merge=lfs -text
16514
+ neuronxcc-2.21.33363.0+82129205/MODULE_248be7993b4edc8d46c8+ac10809c/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
16515
+ neuronxcc-2.21.33363.0+82129205/MODULE_26a5b03803fa3c302f48+ac10809c/model.neff filter=lfs diff=lfs merge=lfs -text
16516
+ neuronxcc-2.21.33363.0+82129205/MODULE_26a5b03803fa3c302f48+ac10809c/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
16517
+ neuronxcc-2.21.33363.0+82129205/MODULE_4c7cbe55e26f80b4ca5e+a32116a7/model.neff filter=lfs diff=lfs merge=lfs -text
16518
+ neuronxcc-2.21.33363.0+82129205/MODULE_4e741b6c6aff91bf0868+ac10809c/model.neff filter=lfs diff=lfs merge=lfs -text
16519
+ neuronxcc-2.21.33363.0+82129205/MODULE_4e741b6c6aff91bf0868+ac10809c/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
16520
+ neuronxcc-2.21.33363.0+82129205/MODULE_5017b58825859b75d379+ac10809c/model.neff filter=lfs diff=lfs merge=lfs -text
16521
+ neuronxcc-2.21.33363.0+82129205/MODULE_5017b58825859b75d379+ac10809c/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
16522
+ neuronxcc-2.21.33363.0+82129205/MODULE_5ffb30cc6ba169632496+ac10809c/model.neff filter=lfs diff=lfs merge=lfs -text
16523
+ neuronxcc-2.21.33363.0+82129205/MODULE_5ffb30cc6ba169632496+ac10809c/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
16524
+ neuronxcc-2.21.33363.0+82129205/MODULE_64c05495132d4c3346af+a32116a7/model.neff filter=lfs diff=lfs merge=lfs -text
16525
+ neuronxcc-2.21.33363.0+82129205/MODULE_688a7baa828603f83864+ac10809c/model.neff filter=lfs diff=lfs merge=lfs -text
16526
+ neuronxcc-2.21.33363.0+82129205/MODULE_688a7baa828603f83864+ac10809c/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
16527
+ neuronxcc-2.21.33363.0+82129205/MODULE_68cb8e9f3c2305481f3a+ac10809c/model.neff filter=lfs diff=lfs merge=lfs -text
16528
+ neuronxcc-2.21.33363.0+82129205/MODULE_68cb8e9f3c2305481f3a+ac10809c/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
16529
+ neuronxcc-2.21.33363.0+82129205/MODULE_749cb1da17f656c57ecd+ac10809c/model.neff filter=lfs diff=lfs merge=lfs -text
16530
+ neuronxcc-2.21.33363.0+82129205/MODULE_749cb1da17f656c57ecd+ac10809c/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
16531
+ neuronxcc-2.21.33363.0+82129205/MODULE_7a50c1d04b1b5328b107+a32116a7/model.neff filter=lfs diff=lfs merge=lfs -text
16532
+ neuronxcc-2.21.33363.0+82129205/MODULE_9b43351b46acfc35f59c+40f75c1a/model.neff filter=lfs diff=lfs merge=lfs -text
16533
+ neuronxcc-2.21.33363.0+82129205/MODULE_9b43351b46acfc35f59c+40f75c1a/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
16534
+ neuronxcc-2.21.33363.0+82129205/MODULE_afeee4236e029e2d644e+a32116a7/model.neff filter=lfs diff=lfs merge=lfs -text
16535
+ neuronxcc-2.21.33363.0+82129205/MODULE_b0b5c42826d33def683b+ac10809c/model.neff filter=lfs diff=lfs merge=lfs -text
16536
+ neuronxcc-2.21.33363.0+82129205/MODULE_b0b5c42826d33def683b+ac10809c/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
16537
+ neuronxcc-2.21.33363.0+82129205/MODULE_b555a5f93b2d6c2b964e+ac10809c/model.neff filter=lfs diff=lfs merge=lfs -text
16538
+ neuronxcc-2.21.33363.0+82129205/MODULE_b555a5f93b2d6c2b964e+ac10809c/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
16539
+ neuronxcc-2.21.33363.0+82129205/MODULE_b61f26bf83def4e7a52c+a32116a7/model.neff filter=lfs diff=lfs merge=lfs -text
16540
+ neuronxcc-2.21.33363.0+82129205/MODULE_b95ab17ef3954bb04026+ac10809c/model.neff filter=lfs diff=lfs merge=lfs -text
16541
+ neuronxcc-2.21.33363.0+82129205/MODULE_b95ab17ef3954bb04026+ac10809c/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
16542
+ neuronxcc-2.21.33363.0+82129205/MODULE_bb55057d1263c7ecc9f1+a32116a7/model.neff filter=lfs diff=lfs merge=lfs -text
16543
+ neuronxcc-2.21.33363.0+82129205/MODULE_bddb314edb25866ecdcb+a32116a7/model.neff filter=lfs diff=lfs merge=lfs -text
16544
+ neuronxcc-2.21.33363.0+82129205/MODULE_be944b6c4fa695088f62+a32116a7/model.neff filter=lfs diff=lfs merge=lfs -text
16545
+ neuronxcc-2.21.33363.0+82129205/MODULE_c0eef7e60449b5b8b913+a32116a7/model.neff filter=lfs diff=lfs merge=lfs -text
16546
+ neuronxcc-2.21.33363.0+82129205/MODULE_c38e6e659433b3e219d8+a32116a7/model.neff filter=lfs diff=lfs merge=lfs -text
16547
+ neuronxcc-2.21.33363.0+82129205/MODULE_c96e8a76473cbbe30609+bf3a5a22/model.neff filter=lfs diff=lfs merge=lfs -text
16548
+ neuronxcc-2.21.33363.0+82129205/MODULE_d130c645df160dbdf55c+bf3a5a22/model.neff filter=lfs diff=lfs merge=lfs -text
16549
+ neuronxcc-2.21.33363.0+82129205/MODULE_db2f47e16eae1f5df700+ac10809c/model.neff filter=lfs diff=lfs merge=lfs -text
16550
+ neuronxcc-2.21.33363.0+82129205/MODULE_db2f47e16eae1f5df700+ac10809c/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
16551
+ neuronxcc-2.21.33363.0+82129205/MODULE_ed645ef81921b62e3ba7+a32116a7/model.neff filter=lfs diff=lfs merge=lfs -text
16552
+ neuronxcc-2.21.33363.0+82129205/MODULE_eea79ad4b5b63ff68d1f+a32116a7/model.neff filter=lfs diff=lfs merge=lfs -text
16553
+ neuronxcc-2.21.33363.0+82129205/MODULE_fffa4c481b28ac601395+ac10809c/model.neff filter=lfs diff=lfs merge=lfs -text
16554
+ neuronxcc-2.21.33363.0+82129205/MODULE_fffa4c481b28ac601395+ac10809c/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/0cc25526e2cfc37a8875a3752f33c4d7505d8a07b869d0f3f41915cf6e763b74/8a4430107c8609f149e2.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": 4,
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.dev2",
39
+ "output_logits": false,
40
+ "pp_degree": 1,
41
+ "sequence_length": 4096,
42
+ "speculation_length": 0,
43
+ "start_rank_id": 0,
44
+ "target": "trn2",
45
+ "torch_dtype": "bfloat16",
46
+ "tp_degree": 4
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.dev2/0cc25526e2cfc37a8875a3752f33c4d7505d8a07b869d0f3f41915cf6e763b74/de33f9f75b62f508e84e.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": 4,
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.dev2",
39
+ "output_logits": false,
40
+ "pp_degree": 1,
41
+ "sequence_length": 8192,
42
+ "speculation_length": 0,
43
+ "start_rank_id": 0,
44
+ "target": "trn2",
45
+ "torch_dtype": "bfloat16",
46
+ "tp_degree": 4
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.dev2/441269935591cad8d370e512c0b93cdd2fce6247c40e5a4866d872ee5338b0de/8d757b039a90a33816e7.json ADDED
@@ -0,0 +1,81 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "unsloth/gemma-3-270m-it",
4
+ "_sliding_window_pattern": 6,
5
+ "_task": "text-generation",
6
+ "architectures": [
7
+ "Gemma3ForCausalLM"
8
+ ],
9
+ "attention_bias": false,
10
+ "attention_dropout": 0.0,
11
+ "attn_logit_softcapping": null,
12
+ "dtype": "bfloat16",
13
+ "final_logit_softcapping": null,
14
+ "head_dim": 256,
15
+ "hidden_activation": "gelu_pytorch_tanh",
16
+ "hidden_size": 640,
17
+ "initializer_range": 0.02,
18
+ "intermediate_size": 2048,
19
+ "layer_types": [
20
+ "sliding_attention",
21
+ "sliding_attention",
22
+ "sliding_attention",
23
+ "sliding_attention",
24
+ "sliding_attention",
25
+ "full_attention",
26
+ "sliding_attention",
27
+ "sliding_attention",
28
+ "sliding_attention",
29
+ "sliding_attention",
30
+ "sliding_attention",
31
+ "full_attention",
32
+ "sliding_attention",
33
+ "sliding_attention",
34
+ "sliding_attention",
35
+ "sliding_attention",
36
+ "sliding_attention",
37
+ "full_attention"
38
+ ],
39
+ "max_position_embeddings": 32768,
40
+ "model_type": "gemma3_text",
41
+ "neuron": {
42
+ "_serialized_key": "NxDNeuronConfig",
43
+ "batch_size": 1,
44
+ "capacity_factor": null,
45
+ "checkpoint_id": "unsloth/gemma-3-270m-it",
46
+ "checkpoint_revision": "23cf460f6bb16954176b3ddcc8d4f250501458a9",
47
+ "continuous_batching": false,
48
+ "ep_degree": 1,
49
+ "fused_qkv": true,
50
+ "glu_mlp": true,
51
+ "local_ranks_size": 4,
52
+ "max_batch_size": 1,
53
+ "max_context_length": 8192,
54
+ "max_topk": 256,
55
+ "n_active_tokens": 8192,
56
+ "neuronxcc_version": "2.21.33363.0+82129205",
57
+ "on_device_sampling": true,
58
+ "optimum_neuron_version": "0.4.5.dev2",
59
+ "output_logits": false,
60
+ "pp_degree": 1,
61
+ "sequence_length": 8192,
62
+ "speculation_length": 0,
63
+ "start_rank_id": 0,
64
+ "target": "trn2",
65
+ "torch_dtype": "bfloat16",
66
+ "tp_degree": 4
67
+ },
68
+ "num_attention_heads": 4,
69
+ "num_hidden_layers": 18,
70
+ "num_key_value_heads": 1,
71
+ "query_pre_attn_scalar": 256,
72
+ "rms_norm_eps": 1e-06,
73
+ "rope_local_base_freq": 10000.0,
74
+ "rope_scaling": null,
75
+ "rope_theta": 1000000.0,
76
+ "sliding_window": 512,
77
+ "unsloth_fixed": true,
78
+ "use_bidirectional_attention": false,
79
+ "use_cache": true,
80
+ "vocab_size": 262144
81
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/4cb7aff9e2a15c151396f2b684013e39d6739f0dec83e5c9dabbfe9d5fcf77b7/287ad99b2a6141ebb12e.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": 4,
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": true,
60
+ "optimum_neuron_version": "0.4.5.dev2",
61
+ "output_logits": false,
62
+ "pp_degree": 1,
63
+ "sequence_length": 4096,
64
+ "speculation_length": 0,
65
+ "start_rank_id": 0,
66
+ "target": "trn2",
67
+ "torch_dtype": "bfloat16",
68
+ "tp_degree": 4
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.dev2/4cb7aff9e2a15c151396f2b684013e39d6739f0dec83e5c9dabbfe9d5fcf77b7/7911379237ca2e32dc36.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": 4,
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.dev2",
61
+ "output_logits": false,
62
+ "pp_degree": 1,
63
+ "sequence_length": 8192,
64
+ "speculation_length": 0,
65
+ "start_rank_id": 0,
66
+ "target": "trn2",
67
+ "torch_dtype": "bfloat16",
68
+ "tp_degree": 4
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.dev2/60feecaa0c4c075e2f3e46a3f55d9a273f0ddd75a0ecf64e4ae27352e0819506/0b9d5658f10ecb783ea7.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "Qwen/Qwen3-Embedding-0.6B",
4
+ "_task": "feature-extraction",
5
+ "architectures": [
6
+ "Qwen3ForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "dtype": "bfloat16",
11
+ "head_dim": 128,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 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": 4,
60
+ "max_batch_size": 4,
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": false,
66
+ "optimum_neuron_version": "0.4.5.dev2",
67
+ "output_logits": false,
68
+ "pp_degree": 1,
69
+ "sequence_length": 8192,
70
+ "speculation_length": 0,
71
+ "start_rank_id": 0,
72
+ "target": "trn2",
73
+ "torch_dtype": "bfloat16",
74
+ "tp_degree": 4
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.dev2/60feecaa0c4c075e2f3e46a3f55d9a273f0ddd75a0ecf64e4ae27352e0819506/3aace2df87913058a524.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "Qwen/Qwen3-Embedding-0.6B",
4
+ "_task": "feature-extraction",
5
+ "architectures": [
6
+ "Qwen3ForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "dtype": "bfloat16",
11
+ "head_dim": 128,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 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": 4,
60
+ "max_batch_size": 6,
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": false,
66
+ "optimum_neuron_version": "0.4.5.dev2",
67
+ "output_logits": false,
68
+ "pp_degree": 1,
69
+ "sequence_length": 8192,
70
+ "speculation_length": 0,
71
+ "start_rank_id": 0,
72
+ "target": "trn2",
73
+ "torch_dtype": "bfloat16",
74
+ "tp_degree": 4
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.dev2/73707b485eab9008c7aba7f5dad0ce2384ac685318d5f888c12fa0d81ed90b19/8ff57fd8c6031855b0ec.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": 4,
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.dev2",
75
+ "output_logits": false,
76
+ "pp_degree": 1,
77
+ "sequence_length": 8192,
78
+ "speculation_length": 0,
79
+ "start_rank_id": 0,
80
+ "target": "trn2",
81
+ "torch_dtype": "bfloat16",
82
+ "tp_degree": 4
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.dev2/73707b485eab9008c7aba7f5dad0ce2384ac685318d5f888c12fa0d81ed90b19/c100f806a6006feef39b.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": 4,
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.dev2",
75
+ "output_logits": false,
76
+ "pp_degree": 1,
77
+ "sequence_length": 4096,
78
+ "speculation_length": 0,
79
+ "start_rank_id": 0,
80
+ "target": "trn2",
81
+ "torch_dtype": "bfloat16",
82
+ "tp_degree": 4
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.dev2/8c90ac2593ed0b7f1ecb60e82cb184fb11f2ea640befa1cc7b10766a5c02525d/3943e7730d04a5ca4a9d.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": 4,
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.dev2",
40
+ "output_logits": false,
41
+ "pp_degree": 1,
42
+ "sequence_length": 4096,
43
+ "speculation_length": 0,
44
+ "start_rank_id": 0,
45
+ "target": "trn2",
46
+ "torch_dtype": "bfloat16",
47
+ "tp_degree": 4
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.dev2/8c90ac2593ed0b7f1ecb60e82cb184fb11f2ea640befa1cc7b10766a5c02525d/fa7402288853e31635c0.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": 4,
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.dev2",
40
+ "output_logits": false,
41
+ "pp_degree": 1,
42
+ "sequence_length": 8192,
43
+ "speculation_length": 0,
44
+ "start_rank_id": 0,
45
+ "target": "trn2",
46
+ "torch_dtype": "bfloat16",
47
+ "tp_degree": 4
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.dev2/929b02754a13cbfdf657d863c3fc6f3bce672879bc6ae48ab45be21e881e9ec2/5b6c82d29f13b10e5052.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": 4,
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.dev2",
67
+ "output_logits": false,
68
+ "pp_degree": 1,
69
+ "sequence_length": 8192,
70
+ "speculation_length": 0,
71
+ "start_rank_id": 0,
72
+ "target": "trn2",
73
+ "torch_dtype": "bfloat16",
74
+ "tp_degree": 4
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.dev2/929b02754a13cbfdf657d863c3fc6f3bce672879bc6ae48ab45be21e881e9ec2/c4069e9e5ecd272fe629.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": 4,
52
+ "capacity_factor": null,
53
+ "checkpoint_id": "Qwen/Qwen3-0.6B",
54
+ "checkpoint_revision": "c1899de289a04d12100db370d81485cdf75e47ca",
55
+ "continuous_batching": true,
56
+ "ep_degree": 1,
57
+ "fused_qkv": true,
58
+ "glu_mlp": true,
59
+ "local_ranks_size": 4,
60
+ "max_batch_size": 4,
61
+ "max_context_length": 4096,
62
+ "max_topk": 256,
63
+ "n_active_tokens": 4096,
64
+ "neuronxcc_version": "2.21.33363.0+82129205",
65
+ "on_device_sampling": true,
66
+ "optimum_neuron_version": "0.4.5.dev2",
67
+ "output_logits": false,
68
+ "pp_degree": 1,
69
+ "sequence_length": 4096,
70
+ "speculation_length": 0,
71
+ "start_rank_id": 0,
72
+ "target": "trn2",
73
+ "torch_dtype": "bfloat16",
74
+ "tp_degree": 4
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.dev2/bd8b5aaeb4693c199f97f6eebd1bbca77255bea6e0ed042c0c592ee5adccec97/262dbf43da8b9b831afa.json ADDED
@@ -0,0 +1,80 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "google/gemma-3-270m-it",
4
+ "_sliding_window_pattern": 6,
5
+ "_task": "text-generation",
6
+ "architectures": [
7
+ "Gemma3ForCausalLM"
8
+ ],
9
+ "attention_bias": false,
10
+ "attention_dropout": 0.0,
11
+ "attn_logit_softcapping": null,
12
+ "dtype": "bfloat16",
13
+ "final_logit_softcapping": null,
14
+ "head_dim": 256,
15
+ "hidden_activation": "gelu_pytorch_tanh",
16
+ "hidden_size": 640,
17
+ "initializer_range": 0.02,
18
+ "intermediate_size": 2048,
19
+ "layer_types": [
20
+ "sliding_attention",
21
+ "sliding_attention",
22
+ "sliding_attention",
23
+ "sliding_attention",
24
+ "sliding_attention",
25
+ "full_attention",
26
+ "sliding_attention",
27
+ "sliding_attention",
28
+ "sliding_attention",
29
+ "sliding_attention",
30
+ "sliding_attention",
31
+ "full_attention",
32
+ "sliding_attention",
33
+ "sliding_attention",
34
+ "sliding_attention",
35
+ "sliding_attention",
36
+ "sliding_attention",
37
+ "full_attention"
38
+ ],
39
+ "max_position_embeddings": 32768,
40
+ "model_type": "gemma3_text",
41
+ "neuron": {
42
+ "_serialized_key": "NxDNeuronConfig",
43
+ "batch_size": 1,
44
+ "capacity_factor": null,
45
+ "checkpoint_id": "google/gemma-3-270m-it",
46
+ "checkpoint_revision": "ac82b4e820549b854eebf28ce6dedaf9fdfa17b3",
47
+ "continuous_batching": false,
48
+ "ep_degree": 1,
49
+ "fused_qkv": true,
50
+ "glu_mlp": true,
51
+ "local_ranks_size": 2,
52
+ "max_batch_size": 1,
53
+ "max_context_length": 1024,
54
+ "max_topk": 256,
55
+ "n_active_tokens": 1024,
56
+ "neuronxcc_version": "2.21.33363.0+82129205",
57
+ "on_device_sampling": true,
58
+ "optimum_neuron_version": "0.4.5.dev2",
59
+ "output_logits": false,
60
+ "pp_degree": 1,
61
+ "sequence_length": 1024,
62
+ "speculation_length": 0,
63
+ "start_rank_id": 0,
64
+ "target": "trn2",
65
+ "torch_dtype": "bfloat16",
66
+ "tp_degree": 2
67
+ },
68
+ "num_attention_heads": 4,
69
+ "num_hidden_layers": 18,
70
+ "num_key_value_heads": 1,
71
+ "query_pre_attn_scalar": 256,
72
+ "rms_norm_eps": 1e-06,
73
+ "rope_local_base_freq": 10000.0,
74
+ "rope_scaling": null,
75
+ "rope_theta": 1000000.0,
76
+ "sliding_window": 512,
77
+ "use_bidirectional_attention": false,
78
+ "use_cache": true,
79
+ "vocab_size": 262144
80
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/bd8b5aaeb4693c199f97f6eebd1bbca77255bea6e0ed042c0c592ee5adccec97/a5218019da46cb49d020.json ADDED
@@ -0,0 +1,80 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "google/gemma-3-270m-it",
4
+ "_sliding_window_pattern": 6,
5
+ "_task": "text-generation",
6
+ "architectures": [
7
+ "Gemma3ForCausalLM"
8
+ ],
9
+ "attention_bias": false,
10
+ "attention_dropout": 0.0,
11
+ "attn_logit_softcapping": null,
12
+ "dtype": "bfloat16",
13
+ "final_logit_softcapping": null,
14
+ "head_dim": 256,
15
+ "hidden_activation": "gelu_pytorch_tanh",
16
+ "hidden_size": 640,
17
+ "initializer_range": 0.02,
18
+ "intermediate_size": 2048,
19
+ "layer_types": [
20
+ "sliding_attention",
21
+ "sliding_attention",
22
+ "sliding_attention",
23
+ "sliding_attention",
24
+ "sliding_attention",
25
+ "full_attention",
26
+ "sliding_attention",
27
+ "sliding_attention",
28
+ "sliding_attention",
29
+ "sliding_attention",
30
+ "sliding_attention",
31
+ "full_attention",
32
+ "sliding_attention",
33
+ "sliding_attention",
34
+ "sliding_attention",
35
+ "sliding_attention",
36
+ "sliding_attention",
37
+ "full_attention"
38
+ ],
39
+ "max_position_embeddings": 32768,
40
+ "model_type": "gemma3_text",
41
+ "neuron": {
42
+ "_serialized_key": "NxDNeuronConfig",
43
+ "batch_size": 1,
44
+ "capacity_factor": null,
45
+ "checkpoint_id": "google/gemma-3-270m-it",
46
+ "checkpoint_revision": "ac82b4e820549b854eebf28ce6dedaf9fdfa17b3",
47
+ "continuous_batching": false,
48
+ "ep_degree": 1,
49
+ "fused_qkv": true,
50
+ "glu_mlp": true,
51
+ "local_ranks_size": 4,
52
+ "max_batch_size": 1,
53
+ "max_context_length": 8192,
54
+ "max_topk": 256,
55
+ "n_active_tokens": 8192,
56
+ "neuronxcc_version": "2.21.33363.0+82129205",
57
+ "on_device_sampling": true,
58
+ "optimum_neuron_version": "0.4.5.dev2",
59
+ "output_logits": false,
60
+ "pp_degree": 1,
61
+ "sequence_length": 8192,
62
+ "speculation_length": 0,
63
+ "start_rank_id": 0,
64
+ "target": "trn2",
65
+ "torch_dtype": "bfloat16",
66
+ "tp_degree": 4
67
+ },
68
+ "num_attention_heads": 4,
69
+ "num_hidden_layers": 18,
70
+ "num_key_value_heads": 1,
71
+ "query_pre_attn_scalar": 256,
72
+ "rms_norm_eps": 1e-06,
73
+ "rope_local_base_freq": 10000.0,
74
+ "rope_scaling": null,
75
+ "rope_theta": 1000000.0,
76
+ "sliding_window": 512,
77
+ "use_bidirectional_attention": false,
78
+ "use_cache": true,
79
+ "vocab_size": 262144
80
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/bd8b5aaeb4693c199f97f6eebd1bbca77255bea6e0ed042c0c592ee5adccec97/c5893ac2150961103e95.json ADDED
@@ -0,0 +1,80 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "google/gemma-3-270m-it",
4
+ "_sliding_window_pattern": 6,
5
+ "_task": "text-generation",
6
+ "architectures": [
7
+ "Gemma3ForCausalLM"
8
+ ],
9
+ "attention_bias": false,
10
+ "attention_dropout": 0.0,
11
+ "attn_logit_softcapping": null,
12
+ "dtype": "bfloat16",
13
+ "final_logit_softcapping": null,
14
+ "head_dim": 256,
15
+ "hidden_activation": "gelu_pytorch_tanh",
16
+ "hidden_size": 640,
17
+ "initializer_range": 0.02,
18
+ "intermediate_size": 2048,
19
+ "layer_types": [
20
+ "sliding_attention",
21
+ "sliding_attention",
22
+ "sliding_attention",
23
+ "sliding_attention",
24
+ "sliding_attention",
25
+ "full_attention",
26
+ "sliding_attention",
27
+ "sliding_attention",
28
+ "sliding_attention",
29
+ "sliding_attention",
30
+ "sliding_attention",
31
+ "full_attention",
32
+ "sliding_attention",
33
+ "sliding_attention",
34
+ "sliding_attention",
35
+ "sliding_attention",
36
+ "sliding_attention",
37
+ "full_attention"
38
+ ],
39
+ "max_position_embeddings": 32768,
40
+ "model_type": "gemma3_text",
41
+ "neuron": {
42
+ "_serialized_key": "NxDNeuronConfig",
43
+ "batch_size": 4,
44
+ "capacity_factor": null,
45
+ "checkpoint_id": "google/gemma-3-270m-it",
46
+ "checkpoint_revision": "ac82b4e820549b854eebf28ce6dedaf9fdfa17b3",
47
+ "continuous_batching": true,
48
+ "ep_degree": 1,
49
+ "fused_qkv": true,
50
+ "glu_mlp": true,
51
+ "local_ranks_size": 4,
52
+ "max_batch_size": 4,
53
+ "max_context_length": 4096,
54
+ "max_topk": 256,
55
+ "n_active_tokens": 4096,
56
+ "neuronxcc_version": "2.21.33363.0+82129205",
57
+ "on_device_sampling": true,
58
+ "optimum_neuron_version": "0.4.5.dev2",
59
+ "output_logits": false,
60
+ "pp_degree": 1,
61
+ "sequence_length": 4096,
62
+ "speculation_length": 0,
63
+ "start_rank_id": 0,
64
+ "target": "trn2",
65
+ "torch_dtype": "bfloat16",
66
+ "tp_degree": 4
67
+ },
68
+ "num_attention_heads": 4,
69
+ "num_hidden_layers": 18,
70
+ "num_key_value_heads": 1,
71
+ "query_pre_attn_scalar": 256,
72
+ "rms_norm_eps": 1e-06,
73
+ "rope_local_base_freq": 10000.0,
74
+ "rope_scaling": null,
75
+ "rope_theta": 1000000.0,
76
+ "sliding_window": 512,
77
+ "use_bidirectional_attention": false,
78
+ "use_cache": true,
79
+ "vocab_size": 262144
80
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/cf6b9a360dcf294104671106bae2adbd9fd291823bb60a351883163684073231/02e21fa4187e1718bddb.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": 4,
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.dev2",
37
+ "output_logits": false,
38
+ "pp_degree": 1,
39
+ "sequence_length": 8192,
40
+ "speculation_length": 0,
41
+ "start_rank_id": 0,
42
+ "target": "trn2",
43
+ "torch_dtype": "bfloat16",
44
+ "tp_degree": 4
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.dev2/cf6b9a360dcf294104671106bae2adbd9fd291823bb60a351883163684073231/b348680583357cdc479f.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": 4,
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.dev2",
37
+ "output_logits": false,
38
+ "pp_degree": 1,
39
+ "sequence_length": 4096,
40
+ "speculation_length": 0,
41
+ "start_rank_id": 0,
42
+ "target": "trn2",
43
+ "torch_dtype": "bfloat16",
44
+ "tp_degree": 4
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.dev2/gemma3_text/google/gemma-3-270m-it/c5893ac2150961103e95.json ADDED
@@ -0,0 +1,80 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "google/gemma-3-270m-it",
4
+ "_sliding_window_pattern": 6,
5
+ "_task": "text-generation",
6
+ "architectures": [
7
+ "Gemma3ForCausalLM"
8
+ ],
9
+ "attention_bias": false,
10
+ "attention_dropout": 0.0,
11
+ "attn_logit_softcapping": null,
12
+ "dtype": "bfloat16",
13
+ "final_logit_softcapping": null,
14
+ "head_dim": 256,
15
+ "hidden_activation": "gelu_pytorch_tanh",
16
+ "hidden_size": 640,
17
+ "initializer_range": 0.02,
18
+ "intermediate_size": 2048,
19
+ "layer_types": [
20
+ "sliding_attention",
21
+ "sliding_attention",
22
+ "sliding_attention",
23
+ "sliding_attention",
24
+ "sliding_attention",
25
+ "full_attention",
26
+ "sliding_attention",
27
+ "sliding_attention",
28
+ "sliding_attention",
29
+ "sliding_attention",
30
+ "sliding_attention",
31
+ "full_attention",
32
+ "sliding_attention",
33
+ "sliding_attention",
34
+ "sliding_attention",
35
+ "sliding_attention",
36
+ "sliding_attention",
37
+ "full_attention"
38
+ ],
39
+ "max_position_embeddings": 32768,
40
+ "model_type": "gemma3_text",
41
+ "neuron": {
42
+ "_serialized_key": "NxDNeuronConfig",
43
+ "batch_size": 4,
44
+ "capacity_factor": null,
45
+ "checkpoint_id": "google/gemma-3-270m-it",
46
+ "checkpoint_revision": "ac82b4e820549b854eebf28ce6dedaf9fdfa17b3",
47
+ "continuous_batching": true,
48
+ "ep_degree": 1,
49
+ "fused_qkv": true,
50
+ "glu_mlp": true,
51
+ "local_ranks_size": 4,
52
+ "max_batch_size": 4,
53
+ "max_context_length": 4096,
54
+ "max_topk": 256,
55
+ "n_active_tokens": 4096,
56
+ "neuronxcc_version": "2.21.33363.0+82129205",
57
+ "on_device_sampling": true,
58
+ "optimum_neuron_version": "0.4.5.dev2",
59
+ "output_logits": false,
60
+ "pp_degree": 1,
61
+ "sequence_length": 4096,
62
+ "speculation_length": 0,
63
+ "start_rank_id": 0,
64
+ "target": "trn2",
65
+ "torch_dtype": "bfloat16",
66
+ "tp_degree": 4
67
+ },
68
+ "num_attention_heads": 4,
69
+ "num_hidden_layers": 18,
70
+ "num_key_value_heads": 1,
71
+ "query_pre_attn_scalar": 256,
72
+ "rms_norm_eps": 1e-06,
73
+ "rope_local_base_freq": 10000.0,
74
+ "rope_scaling": null,
75
+ "rope_theta": 1000000.0,
76
+ "sliding_window": 512,
77
+ "use_bidirectional_attention": false,
78
+ "use_cache": true,
79
+ "vocab_size": 262144
80
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/gemma3_text/unsloth/gemma-3-270m-it/8d757b039a90a33816e7.json ADDED
@@ -0,0 +1,81 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "unsloth/gemma-3-270m-it",
4
+ "_sliding_window_pattern": 6,
5
+ "_task": "text-generation",
6
+ "architectures": [
7
+ "Gemma3ForCausalLM"
8
+ ],
9
+ "attention_bias": false,
10
+ "attention_dropout": 0.0,
11
+ "attn_logit_softcapping": null,
12
+ "dtype": "bfloat16",
13
+ "final_logit_softcapping": null,
14
+ "head_dim": 256,
15
+ "hidden_activation": "gelu_pytorch_tanh",
16
+ "hidden_size": 640,
17
+ "initializer_range": 0.02,
18
+ "intermediate_size": 2048,
19
+ "layer_types": [
20
+ "sliding_attention",
21
+ "sliding_attention",
22
+ "sliding_attention",
23
+ "sliding_attention",
24
+ "sliding_attention",
25
+ "full_attention",
26
+ "sliding_attention",
27
+ "sliding_attention",
28
+ "sliding_attention",
29
+ "sliding_attention",
30
+ "sliding_attention",
31
+ "full_attention",
32
+ "sliding_attention",
33
+ "sliding_attention",
34
+ "sliding_attention",
35
+ "sliding_attention",
36
+ "sliding_attention",
37
+ "full_attention"
38
+ ],
39
+ "max_position_embeddings": 32768,
40
+ "model_type": "gemma3_text",
41
+ "neuron": {
42
+ "_serialized_key": "NxDNeuronConfig",
43
+ "batch_size": 1,
44
+ "capacity_factor": null,
45
+ "checkpoint_id": "unsloth/gemma-3-270m-it",
46
+ "checkpoint_revision": "23cf460f6bb16954176b3ddcc8d4f250501458a9",
47
+ "continuous_batching": false,
48
+ "ep_degree": 1,
49
+ "fused_qkv": true,
50
+ "glu_mlp": true,
51
+ "local_ranks_size": 4,
52
+ "max_batch_size": 1,
53
+ "max_context_length": 8192,
54
+ "max_topk": 256,
55
+ "n_active_tokens": 8192,
56
+ "neuronxcc_version": "2.21.33363.0+82129205",
57
+ "on_device_sampling": true,
58
+ "optimum_neuron_version": "0.4.5.dev2",
59
+ "output_logits": false,
60
+ "pp_degree": 1,
61
+ "sequence_length": 8192,
62
+ "speculation_length": 0,
63
+ "start_rank_id": 0,
64
+ "target": "trn2",
65
+ "torch_dtype": "bfloat16",
66
+ "tp_degree": 4
67
+ },
68
+ "num_attention_heads": 4,
69
+ "num_hidden_layers": 18,
70
+ "num_key_value_heads": 1,
71
+ "query_pre_attn_scalar": 256,
72
+ "rms_norm_eps": 1e-06,
73
+ "rope_local_base_freq": 10000.0,
74
+ "rope_scaling": null,
75
+ "rope_theta": 1000000.0,
76
+ "sliding_window": 512,
77
+ "unsloth_fixed": true,
78
+ "use_bidirectional_attention": false,
79
+ "use_cache": true,
80
+ "vocab_size": 262144
81
+ }
neuronxcc-2.21.33363.0+82129205/0_REGISTRY/0.4.5.dev2/granite/ibm-granite/granite-3.1-2b-instruct/8a4430107c8609f149e2.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": 4,
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.dev2",
39
+ "output_logits": false,
40
+ "pp_degree": 1,
41
+ "sequence_length": 4096,
42
+ "speculation_length": 0,
43
+ "start_rank_id": 0,
44
+ "target": "trn2",
45
+ "torch_dtype": "bfloat16",
46
+ "tp_degree": 4
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.dev2/llama/unsloth/Llama-3.2-1B-Instruct/b348680583357cdc479f.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": 4,
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.dev2",
37
+ "output_logits": false,
38
+ "pp_degree": 1,
39
+ "sequence_length": 4096,
40
+ "speculation_length": 0,
41
+ "start_rank_id": 0,
42
+ "target": "trn2",
43
+ "torch_dtype": "bfloat16",
44
+ "tp_degree": 4
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.dev2/phi3/microsoft/Phi-3.5-mini-instruct/3943e7730d04a5ca4a9d.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": 4,
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.dev2",
40
+ "output_logits": false,
41
+ "pp_degree": 1,
42
+ "sequence_length": 4096,
43
+ "speculation_length": 0,
44
+ "start_rank_id": 0,
45
+ "target": "trn2",
46
+ "torch_dtype": "bfloat16",
47
+ "tp_degree": 4
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.dev2/qwen2/Qwen/Qwen2.5-0.5B/287ad99b2a6141ebb12e.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": 4,
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": true,
60
+ "optimum_neuron_version": "0.4.5.dev2",
61
+ "output_logits": false,
62
+ "pp_degree": 1,
63
+ "sequence_length": 4096,
64
+ "speculation_length": 0,
65
+ "start_rank_id": 0,
66
+ "target": "trn2",
67
+ "torch_dtype": "bfloat16",
68
+ "tp_degree": 4
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.dev2/qwen3/Qwen/Qwen3-0.6B/c4069e9e5ecd272fe629.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": 4,
52
+ "capacity_factor": null,
53
+ "checkpoint_id": "Qwen/Qwen3-0.6B",
54
+ "checkpoint_revision": "c1899de289a04d12100db370d81485cdf75e47ca",
55
+ "continuous_batching": true,
56
+ "ep_degree": 1,
57
+ "fused_qkv": true,
58
+ "glu_mlp": true,
59
+ "local_ranks_size": 4,
60
+ "max_batch_size": 4,
61
+ "max_context_length": 4096,
62
+ "max_topk": 256,
63
+ "n_active_tokens": 4096,
64
+ "neuronxcc_version": "2.21.33363.0+82129205",
65
+ "on_device_sampling": true,
66
+ "optimum_neuron_version": "0.4.5.dev2",
67
+ "output_logits": false,
68
+ "pp_degree": 1,
69
+ "sequence_length": 4096,
70
+ "speculation_length": 0,
71
+ "start_rank_id": 0,
72
+ "target": "trn2",
73
+ "torch_dtype": "bfloat16",
74
+ "tp_degree": 4
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.dev2/qwen3/Qwen/Qwen3-Embedding-0.6B/0b9d5658f10ecb783ea7.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "Qwen/Qwen3-Embedding-0.6B",
4
+ "_task": "feature-extraction",
5
+ "architectures": [
6
+ "Qwen3ForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "dtype": "bfloat16",
11
+ "head_dim": 128,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 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": 4,
60
+ "max_batch_size": 4,
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": false,
66
+ "optimum_neuron_version": "0.4.5.dev2",
67
+ "output_logits": false,
68
+ "pp_degree": 1,
69
+ "sequence_length": 8192,
70
+ "speculation_length": 0,
71
+ "start_rank_id": 0,
72
+ "target": "trn2",
73
+ "torch_dtype": "bfloat16",
74
+ "tp_degree": 4
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.dev2/smollm3/HuggingFaceTB/SmolLM3-3B/c100f806a6006feef39b.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": 4,
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.dev2",
75
+ "output_logits": false,
76
+ "pp_degree": 1,
77
+ "sequence_length": 4096,
78
+ "speculation_length": 0,
79
+ "start_rank_id": 0,
80
+ "target": "trn2",
81
+ "torch_dtype": "bfloat16",
82
+ "tp_degree": 4
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_0e8dfbbde719dcf9a38a+a32116a7/compile_flags.json ADDED
@@ -0,0 +1 @@
 
 
1
+ ["--target=trn2", "--auto-cast=none", "--model-type=transformer", "--tensorizer-options=--enable-ccop-compute-overlap --cc-pipeline-tiling-factor=2 --vectorize-strided-dma ", "-O2", "--lnc=2", "--logfile=/tmp/nxd_model/context_encoding/_tp0_bk0/log-neuron-cc.txt"]
neuronxcc-2.21.33363.0+82129205/MODULE_0e8dfbbde719dcf9a38a+a32116a7/model.done ADDED
File without changes
neuronxcc-2.21.33363.0+82129205/MODULE_0e8dfbbde719dcf9a38a+a32116a7/model.hlo_module.pb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9987727d12a57032344d324fb110355306256fa7cf9ae2f3cdf4defae516a917
3
+ size 2191456
neuronxcc-2.21.33363.0+82129205/MODULE_0e8dfbbde719dcf9a38a+a32116a7/model.neff ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8ca11058a95249aa7afb89bdd1eab8b2e60a25197e2c274d262a678323266bc0
3
+ size 10415104
neuronxcc-2.21.33363.0+82129205/MODULE_164f48a2a338a277ba9b+a32116a7/compile_flags.json ADDED
@@ -0,0 +1 @@
 
 
1
+ ["--target=trn2", "--auto-cast=none", "--model-type=transformer", "--tensorizer-options=--enable-ccop-compute-overlap --cc-pipeline-tiling-factor=2 --vectorize-strided-dma ", "-O2", "--lnc=2", "--logfile=/tmp/nxd_model/context_encoding/_tp0_bk0/log-neuron-cc.txt"]
neuronxcc-2.21.33363.0+82129205/MODULE_164f48a2a338a277ba9b+a32116a7/model.done ADDED
File without changes
neuronxcc-2.21.33363.0+82129205/MODULE_164f48a2a338a277ba9b+a32116a7/model.hlo_module.pb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7ab435dc42282d2ebc9f6e2ce2724fa2b653abebbdd34f2c9f5ac5889d106478
3
+ size 1920963
neuronxcc-2.21.33363.0+82129205/MODULE_164f48a2a338a277ba9b+a32116a7/model.neff ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a702901a3634de3ed8a3a01b4f35401c31eb58d34225c493f5f5d0bd6a75e0d0
3
+ size 35134464
neuronxcc-2.21.33363.0+82129205/MODULE_169cfc2b665743de4e2b+ac10809c/compile_flags.json ADDED
@@ -0,0 +1 @@
 
 
1
+ ["--target=trn2", "--auto-cast=none", "--model-type=transformer", "--tensorizer-options=--enable-ccop-compute-overlap --cc-pipeline-tiling-factor=2 --vectorize-strided-dma ", "-O2", "--lnc=2", "--logfile=/tmp/nxd_model/token_generation/_tp0_bk0/log-neuron-cc.txt", "--enable-internal-neff-wrapper"]
neuronxcc-2.21.33363.0+82129205/MODULE_169cfc2b665743de4e2b+ac10809c/model.done ADDED
File without changes
neuronxcc-2.21.33363.0+82129205/MODULE_169cfc2b665743de4e2b+ac10809c/model.hlo_module.pb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f1fa745a77335f51d1a4f2b0d7217a87f2102be770422b4fff4108abf40be305
3
+ size 1746746
neuronxcc-2.21.33363.0+82129205/MODULE_169cfc2b665743de4e2b+ac10809c/model.neff ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c716830cfeb2d092208e4deb28828bc6fa6bb1c96cb57ee961f52e3c751b472d
3
+ size 2018304
neuronxcc-2.21.33363.0+82129205/MODULE_169cfc2b665743de4e2b+ac10809c/wrapped_neff.hlo ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a5947c5beae4f2e2d7a7b5d403feae2a75e755e99b01ae5add1987d5f80dd182
3
+ size 2174610
neuronxcc-2.21.33363.0+82129205/MODULE_18ad931622c6ffd3a96c+ac10809c/compile_flags.json ADDED
@@ -0,0 +1 @@
 
 
1
+ ["--target=trn2", "--auto-cast=none", "--model-type=transformer", "--tensorizer-options=--enable-ccop-compute-overlap --cc-pipeline-tiling-factor=2 --vectorize-strided-dma ", "-O2", "--lnc=2", "--logfile=/tmp/nxd_model/token_generation/_tp0_bk0/log-neuron-cc.txt", "--enable-internal-neff-wrapper"]
neuronxcc-2.21.33363.0+82129205/MODULE_18ad931622c6ffd3a96c+ac10809c/model.done ADDED
File without changes
neuronxcc-2.21.33363.0+82129205/MODULE_18ad931622c6ffd3a96c+ac10809c/model.hlo_module.pb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ffb2c4769fa4c70ec3f6f2f6a9b4ee47654614a059cf32131080aef740bee558
3
+ size 1946249
neuronxcc-2.21.33363.0+82129205/MODULE_18ad931622c6ffd3a96c+ac10809c/model.neff ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0bfcded84a852029984418e4a5f9457c876961cd473586cf0d2f02ec04014b33
3
+ size 3738624
neuronxcc-2.21.33363.0+82129205/MODULE_18ad931622c6ffd3a96c+ac10809c/wrapped_neff.hlo ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6d7a21ca2e7b095ca207fc77c298f79cdfdae2c584639366ee3224c7cbf2c8ae
3
+ size 3906168
neuronxcc-2.21.33363.0+82129205/MODULE_1d5235b31814f7b0b086+a32116a7/compile_flags.json ADDED
@@ -0,0 +1 @@
 
 
1
+ ["--target=trn2", "--auto-cast=none", "--model-type=transformer", "--tensorizer-options=--enable-ccop-compute-overlap --cc-pipeline-tiling-factor=2 --vectorize-strided-dma ", "-O2", "--lnc=2", "--logfile=/tmp/nxd_model/context_encoding/_tp0_bk0/log-neuron-cc.txt"]
neuronxcc-2.21.33363.0+82129205/MODULE_1d5235b31814f7b0b086+a32116a7/model.done ADDED
File without changes
neuronxcc-2.21.33363.0+82129205/MODULE_1d5235b31814f7b0b086+a32116a7/model.hlo_module.pb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fcc72101146bb93b226e94d352e88db2d1725fcb2a9c431cf17f35e86718f555
3
+ size 1704954
neuronxcc-2.21.33363.0+82129205/MODULE_1d5235b31814f7b0b086+a32116a7/model.neff ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9eaca1c813cd8b22f99d2aa46c641493affc50f5d42b82090d885e5d3816dfbc
3
+ size 17183744