Hugging Face's logo

optimum-internal-testing
/
neuron-testing-cache

Model card Files
xet
Community
dacorvo HF Staff commited on
Commit
af9cfc3
·
verified ·
1 Parent(s): a8b629f

Synchronizing local compiler cache.

Browse files
Files changed (49) hide show
  1. .gitattributes +12 -0
  2. neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.3.dev1/granite/hf-internal-testing/tiny-random-GraniteForCausalLM/c33ed737961273b658ed.json +58 -0
  3. neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.3.dev1/llama/llamafactory/tiny-random-Llama-3/0f11a8a435b5e1600edf.json +62 -0
  4. neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.3.dev1/llama/llamafactory/tiny-random-Llama-3/8b1d4dd2776e8a09feb9.json +62 -0
  5. neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.3.dev1/llama/llamafactory/tiny-random-Llama-3/8dcf1d5215482c17538e.json +62 -0
  6. neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.3.dev1/llama4_text/tiny-random/llama-4/2697e686837ac7a049a8.json +81 -0
  7. neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.3.dev1/mixtral/dacorvo/Mixtral-tiny/1ac71c3e127cabf747eb.json +58 -0
  8. neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.3.dev1/phi3/yujiepan/phi-4-tiny-random/203d827373a64513bf73.json +59 -0
  9. neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.3.dev1/qwen2/yujiepan/qwen2.5-128k-tiny-random/719c46c550ce5344aa92.json +64 -0
  10. neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.3.dev1/qwen3/Qwen/Qwen3-Embedding-0.6B/0b6a12d7e49a9e84983b.json +87 -0
  11. neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.3.dev1/qwen3/Qwen/Qwen3-Embedding-0.6B/c9589527682241ab340e.json +87 -0
  12. neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.3.dev1/qwen3_moe/optimum-internal-testing/tiny-random-qwen3_moe/94702f7d3216c1188b70.json +65 -0
  13. neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.3.dev1/smollm3/HuggingFaceTB/SmolLM3-3B/a1a4920382e3aa2af975.json +134 -0
  14. neuronxcc-2.21.18209.0+043b1bf7/MODULE_1ddf9df4b594957856c3+24129607/compile_flags.json +1 -0
  15. neuronxcc-2.21.18209.0+043b1bf7/MODULE_1ddf9df4b594957856c3+24129607/model.done +0 -0
  16. neuronxcc-2.21.18209.0+043b1bf7/MODULE_1ddf9df4b594957856c3+24129607/model.hlo_module.pb +3 -0
  17. neuronxcc-2.21.18209.0+043b1bf7/MODULE_1ddf9df4b594957856c3+24129607/model.neff +3 -0
  18. neuronxcc-2.21.18209.0+043b1bf7/MODULE_4fdb9fe328288afa6def+ac10809c/compile_flags.json +1 -0
  19. neuronxcc-2.21.18209.0+043b1bf7/MODULE_4fdb9fe328288afa6def+ac10809c/model.done +0 -0
  20. neuronxcc-2.21.18209.0+043b1bf7/MODULE_4fdb9fe328288afa6def+ac10809c/model.hlo_module.pb +3 -0
  21. neuronxcc-2.21.18209.0+043b1bf7/MODULE_4fdb9fe328288afa6def+ac10809c/model.neff +3 -0
  22. neuronxcc-2.21.18209.0+043b1bf7/MODULE_4fdb9fe328288afa6def+ac10809c/wrapped_neff.hlo +3 -0
  23. neuronxcc-2.21.18209.0+043b1bf7/MODULE_91866ed418823bde5ad8+a02c3a36/compile_flags.json +1 -0
  24. neuronxcc-2.21.18209.0+043b1bf7/MODULE_91866ed418823bde5ad8+a02c3a36/model.done +0 -0
  25. neuronxcc-2.21.18209.0+043b1bf7/MODULE_91866ed418823bde5ad8+a02c3a36/model.hlo_module.pb +3 -0
  26. neuronxcc-2.21.18209.0+043b1bf7/MODULE_91866ed418823bde5ad8+a02c3a36/model.neff +3 -0
  27. neuronxcc-2.21.18209.0+043b1bf7/MODULE_91866ed418823bde5ad8+a02c3a36/wrapped_neff.hlo +3 -0
  28. neuronxcc-2.21.18209.0+043b1bf7/MODULE_bc9b13dfd0779a230a2b+65e965ef/compile_flags.json +1 -0
  29. neuronxcc-2.21.18209.0+043b1bf7/MODULE_bc9b13dfd0779a230a2b+65e965ef/model.done +0 -0
  30. neuronxcc-2.21.18209.0+043b1bf7/MODULE_bc9b13dfd0779a230a2b+65e965ef/model.hlo_module.pb +3 -0
  31. neuronxcc-2.21.18209.0+043b1bf7/MODULE_bc9b13dfd0779a230a2b+65e965ef/model.neff +3 -0
  32. neuronxcc-2.21.18209.0+043b1bf7/MODULE_bc9b13dfd0779a230a2b+65e965ef/wrapped_neff.hlo +3 -0
  33. neuronxcc-2.21.18209.0+043b1bf7/MODULE_d7929c32eafd3b893fe9+fb4cc044/compile_flags.json +1 -0
  34. neuronxcc-2.21.18209.0+043b1bf7/MODULE_d7929c32eafd3b893fe9+fb4cc044/model.done +0 -0
  35. neuronxcc-2.21.18209.0+043b1bf7/MODULE_d7929c32eafd3b893fe9+fb4cc044/model.hlo_module.pb +3 -0
  36. neuronxcc-2.21.18209.0+043b1bf7/MODULE_d7929c32eafd3b893fe9+fb4cc044/model.neff +3 -0
  37. neuronxcc-2.21.18209.0+043b1bf7/MODULE_e26c5f9afdbc6d9d4282+a32116a7/compile_flags.json +1 -0
  38. neuronxcc-2.21.18209.0+043b1bf7/MODULE_e26c5f9afdbc6d9d4282+a32116a7/model.done +0 -0
  39. neuronxcc-2.21.18209.0+043b1bf7/MODULE_e26c5f9afdbc6d9d4282+a32116a7/model.hlo_module.pb +3 -0
  40. neuronxcc-2.21.18209.0+043b1bf7/MODULE_e26c5f9afdbc6d9d4282+a32116a7/model.neff +3 -0
  41. neuronxcc-2.21.18209.0+043b1bf7/MODULE_e57c3946dfa9c87cb07f+9936de79/compile_flags.json +1 -0
  42. neuronxcc-2.21.18209.0+043b1bf7/MODULE_e57c3946dfa9c87cb07f+9936de79/model.done +0 -0
  43. neuronxcc-2.21.18209.0+043b1bf7/MODULE_e57c3946dfa9c87cb07f+9936de79/model.hlo_module.pb +3 -0
  44. neuronxcc-2.21.18209.0+043b1bf7/MODULE_e57c3946dfa9c87cb07f+9936de79/model.neff +3 -0
  45. neuronxcc-2.21.18209.0+043b1bf7/MODULE_e57c3946dfa9c87cb07f+9936de79/wrapped_neff.hlo +3 -0
  46. neuronxcc-2.21.18209.0+043b1bf7/MODULE_f9cc02790e4e79aee5af+fb4cc044/compile_flags.json +1 -0
  47. neuronxcc-2.21.18209.0+043b1bf7/MODULE_f9cc02790e4e79aee5af+fb4cc044/model.done +0 -0
  48. neuronxcc-2.21.18209.0+043b1bf7/MODULE_f9cc02790e4e79aee5af+fb4cc044/model.hlo_module.pb +3 -0
  49. neuronxcc-2.21.18209.0+043b1bf7/MODULE_f9cc02790e4e79aee5af+fb4cc044/model.neff +3 -0
.gitattributes CHANGED
@@ -5507,3 +5507,15 @@ neuronxcc-2.21.18209.0+043b1bf7/MODULE_6cb05e0d4a5faa35e069+63e4cd48/model.neff
5507
  neuronxcc-2.21.18209.0+043b1bf7/MODULE_6cb05e0d4a5faa35e069+63e4cd48/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
5508
  neuronxcc-2.21.18209.0+043b1bf7/MODULE_bd37a4fdfc1529582241+cc83d24a/model.neff filter=lfs diff=lfs merge=lfs -text
5509
  neuronxcc-2.21.18209.0+043b1bf7/MODULE_bd37a4fdfc1529582241+cc83d24a/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
 
 
 
 
 
 
 
 
 
 
 
 
 
5507
  neuronxcc-2.21.18209.0+043b1bf7/MODULE_6cb05e0d4a5faa35e069+63e4cd48/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
5508
  neuronxcc-2.21.18209.0+043b1bf7/MODULE_bd37a4fdfc1529582241+cc83d24a/model.neff filter=lfs diff=lfs merge=lfs -text
5509
  neuronxcc-2.21.18209.0+043b1bf7/MODULE_bd37a4fdfc1529582241+cc83d24a/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
5510
+ neuronxcc-2.21.18209.0+043b1bf7/MODULE_1ddf9df4b594957856c3+24129607/model.neff filter=lfs diff=lfs merge=lfs -text
5511
+ neuronxcc-2.21.18209.0+043b1bf7/MODULE_4fdb9fe328288afa6def+ac10809c/model.neff filter=lfs diff=lfs merge=lfs -text
5512
+ neuronxcc-2.21.18209.0+043b1bf7/MODULE_4fdb9fe328288afa6def+ac10809c/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
5513
+ neuronxcc-2.21.18209.0+043b1bf7/MODULE_91866ed418823bde5ad8+a02c3a36/model.neff filter=lfs diff=lfs merge=lfs -text
5514
+ neuronxcc-2.21.18209.0+043b1bf7/MODULE_91866ed418823bde5ad8+a02c3a36/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
5515
+ neuronxcc-2.21.18209.0+043b1bf7/MODULE_bc9b13dfd0779a230a2b+65e965ef/model.neff filter=lfs diff=lfs merge=lfs -text
5516
+ neuronxcc-2.21.18209.0+043b1bf7/MODULE_bc9b13dfd0779a230a2b+65e965ef/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
5517
+ neuronxcc-2.21.18209.0+043b1bf7/MODULE_d7929c32eafd3b893fe9+fb4cc044/model.neff filter=lfs diff=lfs merge=lfs -text
5518
+ neuronxcc-2.21.18209.0+043b1bf7/MODULE_e26c5f9afdbc6d9d4282+a32116a7/model.neff filter=lfs diff=lfs merge=lfs -text
5519
+ neuronxcc-2.21.18209.0+043b1bf7/MODULE_e57c3946dfa9c87cb07f+9936de79/model.neff filter=lfs diff=lfs merge=lfs -text
5520
+ neuronxcc-2.21.18209.0+043b1bf7/MODULE_e57c3946dfa9c87cb07f+9936de79/wrapped_neff.hlo filter=lfs diff=lfs merge=lfs -text
5521
+ neuronxcc-2.21.18209.0+043b1bf7/MODULE_f9cc02790e4e79aee5af+fb4cc044/model.neff filter=lfs diff=lfs merge=lfs -text
neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.3.dev1/granite/hf-internal-testing/tiny-random-GraniteForCausalLM/c33ed737961273b658ed.json ADDED
@@ -0,0 +1,58 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "hf-internal-testing/tiny-random-GraniteForCausalLM",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "GraniteForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "attention_multiplier": 1.0,
11
+ "dtype": "float32",
12
+ "embedding_multiplier": 1.0,
13
+ "hidden_act": "silu",
14
+ "hidden_size": 32,
15
+ "initializer_range": 0.02,
16
+ "intermediate_size": 64,
17
+ "logits_scaling": 1.0,
18
+ "max_position_embeddings": 2048,
19
+ "mlp_bias": false,
20
+ "model_type": "granite",
21
+ "neuron": {
22
+ "_serialized_key": "NxDNeuronConfig",
23
+ "batch_size": 1,
24
+ "capacity_factor": null,
25
+ "checkpoint_id": "hf-internal-testing/tiny-random-GraniteForCausalLM",
26
+ "checkpoint_revision": "c3074ebc0ac2fe545305f5e5f6cce2cc9b2aa0c5",
27
+ "continuous_batching": false,
28
+ "ep_degree": 1,
29
+ "fused_qkv": true,
30
+ "glu_mlp": true,
31
+ "local_ranks_size": 2,
32
+ "max_batch_size": 1,
33
+ "max_context_length": 1024,
34
+ "max_topk": 256,
35
+ "n_active_tokens": 1024,
36
+ "neuronxcc_version": "2.21.18209.0+043b1bf7",
37
+ "on_device_sampling": true,
38
+ "optimum_neuron_version": "0.4.3.dev1",
39
+ "output_logits": false,
40
+ "pp_degree": 1,
41
+ "sequence_length": 1024,
42
+ "speculation_length": 0,
43
+ "start_rank_id": 0,
44
+ "target": "trn1",
45
+ "torch_dtype": "float32",
46
+ "tp_degree": 2
47
+ },
48
+ "num_attention_heads": 4,
49
+ "num_hidden_layers": 2,
50
+ "num_key_value_heads": 4,
51
+ "residual_multiplier": 1.0,
52
+ "rms_norm_eps": 1e-06,
53
+ "rope_scaling": null,
54
+ "rope_theta": 10000.0,
55
+ "tie_word_embeddings": false,
56
+ "use_cache": true,
57
+ "vocab_size": 49152
58
+ }
neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.3.dev1/llama/llamafactory/tiny-random-Llama-3/0f11a8a435b5e1600edf.json ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "llamafactory/tiny-random-Llama-3",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "LlamaForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "dtype": "float16",
11
+ "head_dim": 4,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 16,
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 64,
16
+ "max_position_embeddings": 131072,
17
+ "mlp_bias": false,
18
+ "model_type": "llama",
19
+ "neuron": {
20
+ "_serialized_key": "NxDNeuronConfig",
21
+ "batch_size": 1,
22
+ "capacity_factor": null,
23
+ "checkpoint_id": "llamafactory/tiny-random-Llama-3",
24
+ "checkpoint_revision": "bf2a2e3bf199ad2ee96f02a3c00246c608db22a8",
25
+ "continuous_batching": false,
26
+ "ep_degree": 1,
27
+ "fused_qkv": true,
28
+ "glu_mlp": true,
29
+ "local_ranks_size": 2,
30
+ "max_batch_size": 1,
31
+ "max_context_length": 512,
32
+ "max_topk": 256,
33
+ "n_active_tokens": 512,
34
+ "neuronxcc_version": "2.21.18209.0+043b1bf7",
35
+ "on_device_sampling": true,
36
+ "optimum_neuron_version": "0.4.3.dev1",
37
+ "output_logits": false,
38
+ "pp_degree": 1,
39
+ "sequence_length": 512,
40
+ "speculation_length": 0,
41
+ "start_rank_id": 0,
42
+ "target": "trn1",
43
+ "torch_dtype": "float16",
44
+ "tp_degree": 2
45
+ },
46
+ "num_attention_heads": 4,
47
+ "num_hidden_layers": 2,
48
+ "num_key_value_heads": 4,
49
+ "pretraining_tp": 1,
50
+ "rms_norm_eps": 1e-05,
51
+ "rope_scaling": {
52
+ "factor": 8.0,
53
+ "high_freq_factor": 4.0,
54
+ "low_freq_factor": 1.0,
55
+ "original_max_position_embeddings": 8192,
56
+ "rope_type": "llama3"
57
+ },
58
+ "rope_theta": 500000.0,
59
+ "tie_word_embeddings": false,
60
+ "use_cache": true,
61
+ "vocab_size": 128256
62
+ }
neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.3.dev1/llama/llamafactory/tiny-random-Llama-3/8b1d4dd2776e8a09feb9.json ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "llamafactory/tiny-random-Llama-3",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "LlamaForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "dtype": "float16",
11
+ "head_dim": 4,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 16,
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 64,
16
+ "max_position_embeddings": 131072,
17
+ "mlp_bias": false,
18
+ "model_type": "llama",
19
+ "neuron": {
20
+ "_serialized_key": "NxDNeuronConfig",
21
+ "batch_size": 1,
22
+ "capacity_factor": null,
23
+ "checkpoint_id": "llamafactory/tiny-random-Llama-3",
24
+ "checkpoint_revision": "bf2a2e3bf199ad2ee96f02a3c00246c608db22a8",
25
+ "continuous_batching": false,
26
+ "ep_degree": 1,
27
+ "fused_qkv": true,
28
+ "glu_mlp": true,
29
+ "local_ranks_size": 2,
30
+ "max_batch_size": 1,
31
+ "max_context_length": 1024,
32
+ "max_topk": 256,
33
+ "n_active_tokens": 1024,
34
+ "neuronxcc_version": "2.21.18209.0+043b1bf7",
35
+ "on_device_sampling": true,
36
+ "optimum_neuron_version": "0.4.3.dev1",
37
+ "output_logits": false,
38
+ "pp_degree": 1,
39
+ "sequence_length": 1024,
40
+ "speculation_length": 0,
41
+ "start_rank_id": 0,
42
+ "target": "trn1",
43
+ "torch_dtype": "float16",
44
+ "tp_degree": 2
45
+ },
46
+ "num_attention_heads": 4,
47
+ "num_hidden_layers": 2,
48
+ "num_key_value_heads": 4,
49
+ "pretraining_tp": 1,
50
+ "rms_norm_eps": 1e-05,
51
+ "rope_scaling": {
52
+ "factor": 8.0,
53
+ "high_freq_factor": 4.0,
54
+ "low_freq_factor": 1.0,
55
+ "original_max_position_embeddings": 8192,
56
+ "rope_type": "llama3"
57
+ },
58
+ "rope_theta": 500000.0,
59
+ "tie_word_embeddings": false,
60
+ "use_cache": true,
61
+ "vocab_size": 128256
62
+ }
neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.3.dev1/llama/llamafactory/tiny-random-Llama-3/8dcf1d5215482c17538e.json ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "llamafactory/tiny-random-Llama-3",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "LlamaForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "dtype": "float16",
11
+ "head_dim": 4,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 16,
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 64,
16
+ "max_position_embeddings": 131072,
17
+ "mlp_bias": false,
18
+ "model_type": "llama",
19
+ "neuron": {
20
+ "_serialized_key": "NxDNeuronConfig",
21
+ "batch_size": 2,
22
+ "capacity_factor": null,
23
+ "checkpoint_id": "llamafactory/tiny-random-Llama-3",
24
+ "checkpoint_revision": "bf2a2e3bf199ad2ee96f02a3c00246c608db22a8",
25
+ "continuous_batching": true,
26
+ "ep_degree": 1,
27
+ "fused_qkv": true,
28
+ "glu_mlp": true,
29
+ "local_ranks_size": 2,
30
+ "max_batch_size": 2,
31
+ "max_context_length": 128,
32
+ "max_topk": 256,
33
+ "n_active_tokens": 128,
34
+ "neuronxcc_version": "2.21.18209.0+043b1bf7",
35
+ "on_device_sampling": true,
36
+ "optimum_neuron_version": "0.4.3.dev1",
37
+ "output_logits": false,
38
+ "pp_degree": 1,
39
+ "sequence_length": 128,
40
+ "speculation_length": 0,
41
+ "start_rank_id": 0,
42
+ "target": "trn2",
43
+ "torch_dtype": "float16",
44
+ "tp_degree": 2
45
+ },
46
+ "num_attention_heads": 4,
47
+ "num_hidden_layers": 2,
48
+ "num_key_value_heads": 4,
49
+ "pretraining_tp": 1,
50
+ "rms_norm_eps": 1e-05,
51
+ "rope_scaling": {
52
+ "factor": 8.0,
53
+ "high_freq_factor": 4.0,
54
+ "low_freq_factor": 1.0,
55
+ "original_max_position_embeddings": 8192,
56
+ "rope_type": "llama3"
57
+ },
58
+ "rope_theta": 500000.0,
59
+ "tie_word_embeddings": false,
60
+ "use_cache": true,
61
+ "vocab_size": 128256
62
+ }
neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.3.dev1/llama4_text/tiny-random/llama-4/2697e686837ac7a049a8.json ADDED
@@ -0,0 +1,81 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "tiny-random/llama-4",
4
+ "_task": "text-generation",
5
+ "attention_bias": false,
6
+ "attention_chunk_size": 128,
7
+ "attention_dropout": 0.0,
8
+ "attn_scale": 0.1,
9
+ "attn_temperature_tuning": 4,
10
+ "cache_implementation": "hybrid",
11
+ "dtype": "bfloat16",
12
+ "floor_scale": 8192,
13
+ "for_llm_compressor": false,
14
+ "head_dim": 32,
15
+ "hidden_act": "silu",
16
+ "hidden_size": 32,
17
+ "initializer_range": 0.02,
18
+ "interleave_moe_layer_step": 2,
19
+ "intermediate_size": 64,
20
+ "intermediate_size_mlp": 128,
21
+ "layer_types": [
22
+ "chunked_attention",
23
+ "chunked_attention",
24
+ "chunked_attention",
25
+ "full_attention"
26
+ ],
27
+ "max_position_embeddings": 1048576,
28
+ "model_type": "llama4_text",
29
+ "moe_layers": [
30
+ 1,
31
+ 3
32
+ ],
33
+ "neuron": {
34
+ "_serialized_key": "NxDNeuronConfig",
35
+ "batch_size": 1,
36
+ "capacity_factor": null,
37
+ "checkpoint_id": "tiny-random/llama-4",
38
+ "checkpoint_revision": "9e716f5d4d1ffe0a44a15f46f4a12b840439aba4",
39
+ "continuous_batching": false,
40
+ "ep_degree": 1,
41
+ "fused_qkv": false,
42
+ "glu_mlp": true,
43
+ "local_ranks_size": 2,
44
+ "max_batch_size": 1,
45
+ "max_context_length": 1024,
46
+ "max_topk": 256,
47
+ "n_active_tokens": 1024,
48
+ "neuronxcc_version": "2.21.18209.0+043b1bf7",
49
+ "on_device_sampling": true,
50
+ "optimum_neuron_version": "0.4.3.dev1",
51
+ "output_logits": false,
52
+ "pp_degree": 1,
53
+ "sequence_length": 1024,
54
+ "speculation_length": 0,
55
+ "start_rank_id": 0,
56
+ "target": "trn1",
57
+ "torch_dtype": "bfloat16",
58
+ "tp_degree": 2
59
+ },
60
+ "no_rope_layers": [
61
+ 1,
62
+ 1,
63
+ 1,
64
+ 0
65
+ ],
66
+ "num_attention_heads": 1,
67
+ "num_experts_per_tok": 1,
68
+ "num_hidden_layers": 4,
69
+ "num_key_value_heads": 1,
70
+ "num_local_experts": 8,
71
+ "output_router_logits": false,
72
+ "rms_norm_eps": 1e-05,
73
+ "rope_scaling": null,
74
+ "rope_theta": 500000.0,
75
+ "router_aux_loss_coef": 0.001,
76
+ "router_jitter_noise": 0.0,
77
+ "tie_word_embeddings": true,
78
+ "use_cache": true,
79
+ "use_qk_norm": true,
80
+ "vocab_size": 202048
81
+ }
neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.3.dev1/mixtral/dacorvo/Mixtral-tiny/1ac71c3e127cabf747eb.json ADDED
@@ -0,0 +1,58 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "dacorvo/Mixtral-tiny",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "MixtralForCausalLM"
7
+ ],
8
+ "attention_dropout": 0.0,
9
+ "dtype": "float16",
10
+ "head_dim": 32,
11
+ "hidden_act": "silu",
12
+ "hidden_size": 1024,
13
+ "initializer_range": 0.02,
14
+ "intermediate_size": 3584,
15
+ "max_position_embeddings": 1024,
16
+ "model_type": "mixtral",
17
+ "neuron": {
18
+ "_serialized_key": "NxDNeuronConfig",
19
+ "batch_size": 1,
20
+ "capacity_factor": null,
21
+ "checkpoint_id": "dacorvo/Mixtral-tiny",
22
+ "checkpoint_revision": "c557ba205ddff6ea911f4719e0d543d6c08356b6",
23
+ "continuous_batching": false,
24
+ "ep_degree": 1,
25
+ "fused_qkv": false,
26
+ "glu_mlp": true,
27
+ "local_ranks_size": 2,
28
+ "max_batch_size": 1,
29
+ "max_context_length": 1024,
30
+ "max_topk": 256,
31
+ "n_active_tokens": 1024,
32
+ "neuronxcc_version": "2.21.18209.0+043b1bf7",
33
+ "on_device_sampling": false,
34
+ "optimum_neuron_version": "0.4.3.dev1",
35
+ "output_logits": false,
36
+ "pp_degree": 1,
37
+ "sequence_length": 1024,
38
+ "speculation_length": 0,
39
+ "start_rank_id": 0,
40
+ "target": "trn1",
41
+ "torch_dtype": "float16",
42
+ "tp_degree": 2
43
+ },
44
+ "num_attention_heads": 32,
45
+ "num_experts_per_tok": 2,
46
+ "num_hidden_layers": 2,
47
+ "num_key_value_heads": 8,
48
+ "num_local_experts": 8,
49
+ "output_router_logits": false,
50
+ "rms_norm_eps": 1e-05,
51
+ "rope_theta": 10000.0,
52
+ "router_aux_loss_coef": 0.001,
53
+ "router_jitter_noise": 0.0,
54
+ "sliding_window": 4096,
55
+ "tie_word_embeddings": false,
56
+ "use_cache": true,
57
+ "vocab_size": 32000
58
+ }
neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.3.dev1/phi3/yujiepan/phi-4-tiny-random/203d827373a64513bf73.json ADDED
@@ -0,0 +1,59 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "yujiepan/phi-4-tiny-random",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "Phi3ForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "auto_map": {},
11
+ "dtype": "bfloat16",
12
+ "embd_pdrop": 0.0,
13
+ "hidden_act": "silu",
14
+ "hidden_size": 16,
15
+ "initializer_range": 0.02,
16
+ "intermediate_size": 32,
17
+ "max_position_embeddings": 16384,
18
+ "model_type": "phi3",
19
+ "neuron": {
20
+ "_serialized_key": "NxDNeuronConfig",
21
+ "batch_size": 1,
22
+ "capacity_factor": null,
23
+ "checkpoint_id": "yujiepan/phi-4-tiny-random",
24
+ "checkpoint_revision": "18a9a1168dc97ac6d128f811925670c275610f5a",
25
+ "continuous_batching": false,
26
+ "ep_degree": 1,
27
+ "fused_qkv": true,
28
+ "glu_mlp": true,
29
+ "local_ranks_size": 2,
30
+ "max_batch_size": 1,
31
+ "max_context_length": 1024,
32
+ "max_topk": 256,
33
+ "n_active_tokens": 1024,
34
+ "neuronxcc_version": "2.21.18209.0+043b1bf7",
35
+ "on_device_sampling": true,
36
+ "optimum_neuron_version": "0.4.3.dev1",
37
+ "output_logits": false,
38
+ "pp_degree": 1,
39
+ "sequence_length": 1024,
40
+ "speculation_length": 0,
41
+ "start_rank_id": 0,
42
+ "target": "trn1",
43
+ "torch_dtype": "bfloat16",
44
+ "tp_degree": 2
45
+ },
46
+ "num_attention_heads": 2,
47
+ "num_hidden_layers": 2,
48
+ "num_key_value_heads": 1,
49
+ "original_max_position_embeddings": 16384,
50
+ "partial_rotary_factor": 1.0,
51
+ "resid_pdrop": 0.0,
52
+ "rms_norm_eps": 1e-05,
53
+ "rope_scaling": null,
54
+ "rope_theta": 250000,
55
+ "sliding_window": null,
56
+ "tie_word_embeddings": false,
57
+ "use_cache": true,
58
+ "vocab_size": 100352
59
+ }
neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.3.dev1/qwen2/yujiepan/qwen2.5-128k-tiny-random/719c46c550ce5344aa92.json ADDED
@@ -0,0 +1,64 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "yujiepan/qwen2.5-128k-tiny-random",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "Qwen2ForCausalLM"
7
+ ],
8
+ "attention_dropout": 0.0,
9
+ "dtype": "bfloat16",
10
+ "hidden_act": "silu",
11
+ "hidden_size": 8,
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 16,
14
+ "layer_types": [
15
+ "full_attention",
16
+ "full_attention"
17
+ ],
18
+ "max_position_embeddings": 32768,
19
+ "max_window_layers": 1,
20
+ "model_type": "qwen2",
21
+ "neuron": {
22
+ "_serialized_key": "NxDNeuronConfig",
23
+ "batch_size": 1,
24
+ "capacity_factor": null,
25
+ "checkpoint_id": "yujiepan/qwen2.5-128k-tiny-random",
26
+ "checkpoint_revision": "c8296d4ca3f87782876d2382fbb6481d1beb8ef0",
27
+ "continuous_batching": false,
28
+ "ep_degree": 1,
29
+ "fused_qkv": false,
30
+ "glu_mlp": true,
31
+ "local_ranks_size": 2,
32
+ "max_batch_size": 1,
33
+ "max_context_length": 1024,
34
+ "max_topk": 256,
35
+ "n_active_tokens": 1024,
36
+ "neuronxcc_version": "2.21.18209.0+043b1bf7",
37
+ "on_device_sampling": true,
38
+ "optimum_neuron_version": "0.4.3.dev1",
39
+ "output_logits": false,
40
+ "pp_degree": 1,
41
+ "sequence_length": 1024,
42
+ "speculation_length": 0,
43
+ "start_rank_id": 0,
44
+ "target": "trn1",
45
+ "torch_dtype": "bfloat16",
46
+ "tp_degree": 2
47
+ },
48
+ "num_attention_heads": 4,
49
+ "num_hidden_layers": 2,
50
+ "num_key_value_heads": 2,
51
+ "rms_norm_eps": 1e-06,
52
+ "rope_scaling": {
53
+ "factor": 4.0,
54
+ "original_max_position_embeddings": 32768,
55
+ "rope_type": "yarn",
56
+ "type": "yarn"
57
+ },
58
+ "rope_theta": 1000000.0,
59
+ "sliding_window": null,
60
+ "tie_word_embeddings": false,
61
+ "use_cache": true,
62
+ "use_sliding_window": false,
63
+ "vocab_size": 152064
64
+ }
neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.3.dev1/qwen3/Qwen/Qwen3-Embedding-0.6B/0b6a12d7e49a9e84983b.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "Qwen/Qwen3-Embedding-0.6B",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "Qwen3ForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "dtype": "bfloat16",
11
+ "head_dim": 128,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 1024,
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 3072,
16
+ "layer_types": [
17
+ "full_attention",
18
+ "full_attention",
19
+ "full_attention",
20
+ "full_attention",
21
+ "full_attention",
22
+ "full_attention",
23
+ "full_attention",
24
+ "full_attention",
25
+ "full_attention",
26
+ "full_attention",
27
+ "full_attention",
28
+ "full_attention",
29
+ "full_attention",
30
+ "full_attention",
31
+ "full_attention",
32
+ "full_attention",
33
+ "full_attention",
34
+ "full_attention",
35
+ "full_attention",
36
+ "full_attention",
37
+ "full_attention",
38
+ "full_attention",
39
+ "full_attention",
40
+ "full_attention",
41
+ "full_attention",
42
+ "full_attention",
43
+ "full_attention",
44
+ "full_attention"
45
+ ],
46
+ "max_position_embeddings": 32768,
47
+ "max_window_layers": 28,
48
+ "model_type": "qwen3",
49
+ "neuron": {
50
+ "_serialized_key": "NxDNeuronConfig",
51
+ "batch_size": 6,
52
+ "capacity_factor": null,
53
+ "checkpoint_id": "Qwen/Qwen3-Embedding-0.6B",
54
+ "checkpoint_revision": "c54f2e6e80b2d7b7de06f51cec4959f6b3e03418",
55
+ "continuous_batching": false,
56
+ "ep_degree": 1,
57
+ "fused_qkv": true,
58
+ "glu_mlp": true,
59
+ "local_ranks_size": 2,
60
+ "max_batch_size": 6,
61
+ "max_context_length": 1024,
62
+ "max_topk": 256,
63
+ "n_active_tokens": 1024,
64
+ "neuronxcc_version": "2.21.18209.0+043b1bf7",
65
+ "on_device_sampling": false,
66
+ "optimum_neuron_version": "0.4.3.dev1",
67
+ "output_logits": false,
68
+ "pp_degree": 1,
69
+ "sequence_length": 1024,
70
+ "speculation_length": 0,
71
+ "start_rank_id": 0,
72
+ "target": "trn1",
73
+ "torch_dtype": "bfloat16",
74
+ "tp_degree": 2
75
+ },
76
+ "num_attention_heads": 16,
77
+ "num_hidden_layers": 28,
78
+ "num_key_value_heads": 8,
79
+ "rms_norm_eps": 1e-06,
80
+ "rope_scaling": null,
81
+ "rope_theta": 1000000,
82
+ "sliding_window": null,
83
+ "tie_word_embeddings": true,
84
+ "use_cache": true,
85
+ "use_sliding_window": false,
86
+ "vocab_size": 151669
87
+ }
neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.3.dev1/qwen3/Qwen/Qwen3-Embedding-0.6B/c9589527682241ab340e.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "Qwen/Qwen3-Embedding-0.6B",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "Qwen3ForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "dtype": "bfloat16",
11
+ "head_dim": 128,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 1024,
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 3072,
16
+ "layer_types": [
17
+ "full_attention",
18
+ "full_attention",
19
+ "full_attention",
20
+ "full_attention",
21
+ "full_attention",
22
+ "full_attention",
23
+ "full_attention",
24
+ "full_attention",
25
+ "full_attention",
26
+ "full_attention",
27
+ "full_attention",
28
+ "full_attention",
29
+ "full_attention",
30
+ "full_attention",
31
+ "full_attention",
32
+ "full_attention",
33
+ "full_attention",
34
+ "full_attention",
35
+ "full_attention",
36
+ "full_attention",
37
+ "full_attention",
38
+ "full_attention",
39
+ "full_attention",
40
+ "full_attention",
41
+ "full_attention",
42
+ "full_attention",
43
+ "full_attention",
44
+ "full_attention"
45
+ ],
46
+ "max_position_embeddings": 32768,
47
+ "max_window_layers": 28,
48
+ "model_type": "qwen3",
49
+ "neuron": {
50
+ "_serialized_key": "NxDNeuronConfig",
51
+ "batch_size": 4,
52
+ "capacity_factor": null,
53
+ "checkpoint_id": "Qwen/Qwen3-Embedding-0.6B",
54
+ "checkpoint_revision": "c54f2e6e80b2d7b7de06f51cec4959f6b3e03418",
55
+ "continuous_batching": false,
56
+ "ep_degree": 1,
57
+ "fused_qkv": true,
58
+ "glu_mlp": true,
59
+ "local_ranks_size": 2,
60
+ "max_batch_size": 4,
61
+ "max_context_length": 1024,
62
+ "max_topk": 256,
63
+ "n_active_tokens": 1024,
64
+ "neuronxcc_version": "2.21.18209.0+043b1bf7",
65
+ "on_device_sampling": false,
66
+ "optimum_neuron_version": "0.4.3.dev1",
67
+ "output_logits": false,
68
+ "pp_degree": 1,
69
+ "sequence_length": 1024,
70
+ "speculation_length": 0,
71
+ "start_rank_id": 0,
72
+ "target": "trn1",
73
+ "torch_dtype": "bfloat16",
74
+ "tp_degree": 2
75
+ },
76
+ "num_attention_heads": 16,
77
+ "num_hidden_layers": 28,
78
+ "num_key_value_heads": 8,
79
+ "rms_norm_eps": 1e-06,
80
+ "rope_scaling": null,
81
+ "rope_theta": 1000000,
82
+ "sliding_window": null,
83
+ "tie_word_embeddings": true,
84
+ "use_cache": true,
85
+ "use_sliding_window": false,
86
+ "vocab_size": 151669
87
+ }
neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.3.dev1/qwen3_moe/optimum-internal-testing/tiny-random-qwen3_moe/94702f7d3216c1188b70.json ADDED
@@ -0,0 +1,65 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "optimum-internal-testing/tiny-random-qwen3_moe",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "Qwen3MoeForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "decoder_sparse_step": 2,
11
+ "dtype": "float32",
12
+ "head_dim": 32,
13
+ "hidden_act": "silu",
14
+ "hidden_size": 64,
15
+ "initializer_range": 0.02,
16
+ "intermediate_size": 128,
17
+ "max_position_embeddings": 40960,
18
+ "max_window_layers": 1,
19
+ "mlp_only_layers": [],
20
+ "model_type": "qwen3_moe",
21
+ "moe_intermediate_size": 128,
22
+ "neuron": {
23
+ "_serialized_key": "NxDNeuronConfig",
24
+ "batch_size": 1,
25
+ "capacity_factor": null,
26
+ "checkpoint_id": "optimum-internal-testing/tiny-random-qwen3_moe",
27
+ "checkpoint_revision": "e0230be2839556b44b7400a233c73c74b4abb7af",
28
+ "continuous_batching": false,
29
+ "ep_degree": 1,
30
+ "fused_qkv": false,
31
+ "glu_mlp": true,
32
+ "local_ranks_size": 2,
33
+ "max_batch_size": 1,
34
+ "max_context_length": 1024,
35
+ "max_topk": 256,
36
+ "n_active_tokens": 1024,
37
+ "neuronxcc_version": "2.21.18209.0+043b1bf7",
38
+ "on_device_sampling": true,
39
+ "optimum_neuron_version": "0.4.3.dev1",
40
+ "output_logits": false,
41
+ "pp_degree": 1,
42
+ "sequence_length": 1024,
43
+ "speculation_length": 0,
44
+ "start_rank_id": 0,
45
+ "target": "trn1",
46
+ "torch_dtype": "float32",
47
+ "tp_degree": 2
48
+ },
49
+ "norm_topk_prob": true,
50
+ "num_attention_heads": 2,
51
+ "num_experts": 8,
52
+ "num_experts_per_tok": 2,
53
+ "num_hidden_layers": 2,
54
+ "num_key_value_heads": 1,
55
+ "output_router_logits": false,
56
+ "rms_norm_eps": 1e-06,
57
+ "rope_scaling": null,
58
+ "rope_theta": 1000000.0,
59
+ "router_aux_loss_coef": 0.001,
60
+ "sliding_window": null,
61
+ "tie_word_embeddings": true,
62
+ "use_cache": true,
63
+ "use_sliding_window": false,
64
+ "vocab_size": 151936
65
+ }
neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.3.dev1/smollm3/HuggingFaceTB/SmolLM3-3B/a1a4920382e3aa2af975.json ADDED
@@ -0,0 +1,134 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "HuggingFaceTB/SmolLM3-3B",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "SmolLM3ForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "dtype": "bfloat16",
11
+ "hidden_act": "silu",
12
+ "hidden_size": 2048,
13
+ "initializer_range": 0.02,
14
+ "intermediate_size": 11008,
15
+ "layer_types": [
16
+ "full_attention",
17
+ "full_attention",
18
+ "full_attention",
19
+ "full_attention",
20
+ "full_attention",
21
+ "full_attention",
22
+ "full_attention",
23
+ "full_attention",
24
+ "full_attention",
25
+ "full_attention",
26
+ "full_attention",
27
+ "full_attention",
28
+ "full_attention",
29
+ "full_attention",
30
+ "full_attention",
31
+ "full_attention",
32
+ "full_attention",
33
+ "full_attention",
34
+ "full_attention",
35
+ "full_attention",
36
+ "full_attention",
37
+ "full_attention",
38
+ "full_attention",
39
+ "full_attention",
40
+ "full_attention",
41
+ "full_attention",
42
+ "full_attention",
43
+ "full_attention",
44
+ "full_attention",
45
+ "full_attention",
46
+ "full_attention",
47
+ "full_attention",
48
+ "full_attention",
49
+ "full_attention",
50
+ "full_attention",
51
+ "full_attention"
52
+ ],
53
+ "max_position_embeddings": 65536,
54
+ "max_window_layers": 28,
55
+ "mlp_bias": false,
56
+ "model_type": "smollm3",
57
+ "neuron": {
58
+ "_serialized_key": "NxDNeuronConfig",
59
+ "batch_size": 1,
60
+ "capacity_factor": null,
61
+ "checkpoint_id": "HuggingFaceTB/SmolLM3-3B",
62
+ "checkpoint_revision": "a07cc9a04f16550a088caea529712d1d335b0ac1",
63
+ "continuous_batching": false,
64
+ "ep_degree": 1,
65
+ "fused_qkv": true,
66
+ "glu_mlp": true,
67
+ "local_ranks_size": 2,
68
+ "max_batch_size": 1,
69
+ "max_context_length": 1024,
70
+ "max_topk": 256,
71
+ "n_active_tokens": 1024,
72
+ "neuronxcc_version": "2.21.18209.0+043b1bf7",
73
+ "on_device_sampling": true,
74
+ "optimum_neuron_version": "0.4.3.dev1",
75
+ "output_logits": false,
76
+ "pp_degree": 1,
77
+ "sequence_length": 1024,
78
+ "speculation_length": 0,
79
+ "start_rank_id": 0,
80
+ "target": "trn1",
81
+ "torch_dtype": "bfloat16",
82
+ "tp_degree": 2
83
+ },
84
+ "no_rope_layer_interval": 4,
85
+ "no_rope_layers": [
86
+ 1,
87
+ 1,
88
+ 1,
89
+ 0,
90
+ 1,
91
+ 1,
92
+ 1,
93
+ 0,
94
+ 1,
95
+ 1,
96
+ 1,
97
+ 0,
98
+ 1,
99
+ 1,
100
+ 1,
101
+ 0,
102
+ 1,
103
+ 1,
104
+ 1,
105
+ 0,
106
+ 1,
107
+ 1,
108
+ 1,
109
+ 0,
110
+ 1,
111
+ 1,
112
+ 1,
113
+ 0,
114
+ 1,
115
+ 1,
116
+ 1,
117
+ 0,
118
+ 1,
119
+ 1,
120
+ 1,
121
+ 0
122
+ ],
123
+ "num_attention_heads": 16,
124
+ "num_hidden_layers": 36,
125
+ "num_key_value_heads": 4,
126
+ "pretraining_tp": 2,
127
+ "rms_norm_eps": 1e-06,
128
+ "rope_scaling": null,
129
+ "rope_theta": 5000000.0,
130
+ "sliding_window": null,
131
+ "use_cache": false,
132
+ "use_sliding_window": false,
133
+ "vocab_size": 128256
134
+ }
neuronxcc-2.21.18209.0+043b1bf7/MODULE_1ddf9df4b594957856c3+24129607/compile_flags.json ADDED
@@ -0,0 +1 @@
 
 
1
+ ["--target=trn1", "--auto-cast=none", "--model-type=transformer", "--tensorizer-options=--enable-ccop-compute-overlap --cc-pipeline-tiling-factor=2 --vectorize-strided-dma ", "-O2", "--lnc=1", "--logfile=/tmp/nxd_model/context_encoding/_tp0_bk0/log-neuron-cc.txt"]
neuronxcc-2.21.18209.0+043b1bf7/MODULE_1ddf9df4b594957856c3+24129607/model.done ADDED
File without changes
neuronxcc-2.21.18209.0+043b1bf7/MODULE_1ddf9df4b594957856c3+24129607/model.hlo_module.pb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:01ab37542d3cf3b6441fb14fa043deb6d221000e4cc9894128080364cedc9963
3
+ size 83504
neuronxcc-2.21.18209.0+043b1bf7/MODULE_1ddf9df4b594957856c3+24129607/model.neff ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b77f3686480f7953a4668d9232a48ba38d31481bc8cc4e99a4d4ad88d9cfac7f
3
+ size 287744
neuronxcc-2.21.18209.0+043b1bf7/MODULE_4fdb9fe328288afa6def+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-2.21.18209.0+043b1bf7/MODULE_4fdb9fe328288afa6def+ac10809c/model.done ADDED
File without changes
neuronxcc-2.21.18209.0+043b1bf7/MODULE_4fdb9fe328288afa6def+ac10809c/model.hlo_module.pb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c5d477f34ae752f5a59c170824b8b4808870f92b137ec5c4cf7cb18299e7a0c7
3
+ size 1574268
neuronxcc-2.21.18209.0+043b1bf7/MODULE_4fdb9fe328288afa6def+ac10809c/model.neff ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:61419326ab66344bb85210ccd91a2dabc43b1e9d97203480a85ae50ddda77c04
3
+ size 779264
neuronxcc-2.21.18209.0+043b1bf7/MODULE_4fdb9fe328288afa6def+ac10809c/wrapped_neff.hlo ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:755d6145b88a566bf58b635ca8a2acd615b4987d1523f3bbc2c377ceac06fc67
3
+ size 787211
neuronxcc-2.21.18209.0+043b1bf7/MODULE_91866ed418823bde5ad8+a02c3a36/compile_flags.json ADDED
@@ -0,0 +1 @@
 
 
1
+ ["--target=trn1", "--auto-cast=none", "--model-type=transformer", "--tensorizer-options=--enable-ccop-compute-overlap --cc-pipeline-tiling-factor=2 --vectorize-strided-dma ", "-O2", "--lnc=1", "--logfile=/tmp/nxd_model/token_generation/_tp0_bk0/log-neuron-cc.txt", "--enable-internal-neff-wrapper"]
neuronxcc-2.21.18209.0+043b1bf7/MODULE_91866ed418823bde5ad8+a02c3a36/model.done ADDED
File without changes
neuronxcc-2.21.18209.0+043b1bf7/MODULE_91866ed418823bde5ad8+a02c3a36/model.hlo_module.pb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ac09fee9beef4356d5fa3b7db4c5cc47c086e3b3d0c0d3bd4b31eb45baf9c88a
3
+ size 82753
neuronxcc-2.21.18209.0+043b1bf7/MODULE_91866ed418823bde5ad8+a02c3a36/model.neff ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:80cae2bfc2f488313e13f34a29c65602fcbefdc238f23694a0b5bcbc1af1d422
3
+ size 277504
neuronxcc-2.21.18209.0+043b1bf7/MODULE_91866ed418823bde5ad8+a02c3a36/wrapped_neff.hlo ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:def8afe6cacee56b3c29283920281638bd10aff127c441ae7fb7eacda0856d8c
3
+ size 285854
neuronxcc-2.21.18209.0+043b1bf7/MODULE_bc9b13dfd0779a230a2b+65e965ef/compile_flags.json ADDED
@@ -0,0 +1 @@
 
 
1
+ ["--target=trn1", "--enable-saturate-infinity", "--auto-cast=none", "--model-type=transformer", "-O1", "--logfile=/tmp/nxdi_test_f01d4732-5b15-4069-94f7-6be6413e7ca3/compiler_workdir/AttentionModelWrapper/_tp0_bk0/log-neuron-cc.txt", "--enable-internal-neff-wrapper"]
neuronxcc-2.21.18209.0+043b1bf7/MODULE_bc9b13dfd0779a230a2b+65e965ef/model.done ADDED
File without changes
neuronxcc-2.21.18209.0+043b1bf7/MODULE_bc9b13dfd0779a230a2b+65e965ef/model.hlo_module.pb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4590b82eb167fd46963abdbb717cc18cb9511961a3393d100eeacdbdc45b0243
3
+ size 11280
neuronxcc-2.21.18209.0+043b1bf7/MODULE_bc9b13dfd0779a230a2b+65e965ef/model.neff ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dd7ce849c612a7ed9ea3ac7d64a06917f2c08a43682d4f72d905366ac68abb71
3
+ size 1444864
neuronxcc-2.21.18209.0+043b1bf7/MODULE_bc9b13dfd0779a230a2b+65e965ef/wrapped_neff.hlo ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:da2f6164507c66cc56b5530ddfb2863127fe0100587bd87fee4529e3f9acbac0
3
+ size 1447845
neuronxcc-2.21.18209.0+043b1bf7/MODULE_d7929c32eafd3b893fe9+fb4cc044/compile_flags.json ADDED
@@ -0,0 +1 @@
 
 
1
+ ["--target=trn1", "--auto-cast=none", "--model-type=transformer", "--tensorizer-options=--enable-ccop-compute-overlap --cc-pipeline-tiling-factor=2 --vectorize-strided-dma ", "-O2", "--lnc=1", "--logfile=/tmp/nxd_model/encoding/_tp0_bk0/log-neuron-cc.txt"]
neuronxcc-2.21.18209.0+043b1bf7/MODULE_d7929c32eafd3b893fe9+fb4cc044/model.done ADDED
File without changes
neuronxcc-2.21.18209.0+043b1bf7/MODULE_d7929c32eafd3b893fe9+fb4cc044/model.hlo_module.pb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ddb0f3286fa552a5deaedba2813deba9d73bbc5320fd992fe39d99e5f3cca787
3
+ size 490318
neuronxcc-2.21.18209.0+043b1bf7/MODULE_d7929c32eafd3b893fe9+fb4cc044/model.neff ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5550253bfa791e4f5e329e17f417c0c8ebef7afeb40e1b887373673569b8a961
3
+ size 27905024
neuronxcc-2.21.18209.0+043b1bf7/MODULE_e26c5f9afdbc6d9d4282+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-2.21.18209.0+043b1bf7/MODULE_e26c5f9afdbc6d9d4282+a32116a7/model.done ADDED
File without changes
neuronxcc-2.21.18209.0+043b1bf7/MODULE_e26c5f9afdbc6d9d4282+a32116a7/model.hlo_module.pb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:92def4bd5355f728fc79803e92305cfb7b33bcadb93eb1c52a4ff93e6a75c216
3
+ size 1289358
neuronxcc-2.21.18209.0+043b1bf7/MODULE_e26c5f9afdbc6d9d4282+a32116a7/model.neff ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7f3b9a30250da9baba26f1a714b3859f100f1fb9dcc7649f26580db9221f3510
3
+ size 728064
neuronxcc-2.21.18209.0+043b1bf7/MODULE_e57c3946dfa9c87cb07f+9936de79/compile_flags.json ADDED
@@ -0,0 +1 @@
 
 
1
+ ["--target=trn1", "--enable-saturate-infinity", "--auto-cast=none", "--model-type=transformer", "-O1", "--logfile=/tmp/nxdi_test_3b92753c-e0ff-4aea-8057-8b2cc32e8e5a/compiler_workdir/AttentionModelWrapper/_tp0_bk0/log-neuron-cc.txt", "--enable-internal-neff-wrapper"]
neuronxcc-2.21.18209.0+043b1bf7/MODULE_e57c3946dfa9c87cb07f+9936de79/model.done ADDED
File without changes
neuronxcc-2.21.18209.0+043b1bf7/MODULE_e57c3946dfa9c87cb07f+9936de79/model.hlo_module.pb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:51e79a5cc325f9a317d158d931d75ec73ae1cfb16a4d83343a2593edd95d0044
3
+ size 8979
neuronxcc-2.21.18209.0+043b1bf7/MODULE_e57c3946dfa9c87cb07f+9936de79/model.neff ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a53da14e04f984cfa1d1934207c8353adc2143392365770f700f869b59257c13
3
+ size 246784
neuronxcc-2.21.18209.0+043b1bf7/MODULE_e57c3946dfa9c87cb07f+9936de79/wrapped_neff.hlo ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f5ce6725f9e801fa5163ab54d3124e4a1e82c04a70cb3fab8807b431ef85f54e
3
+ size 249608
neuronxcc-2.21.18209.0+043b1bf7/MODULE_f9cc02790e4e79aee5af+fb4cc044/compile_flags.json ADDED
@@ -0,0 +1 @@
 
 
1
+ ["--target=trn1", "--auto-cast=none", "--model-type=transformer", "--tensorizer-options=--enable-ccop-compute-overlap --cc-pipeline-tiling-factor=2 --vectorize-strided-dma ", "-O2", "--lnc=1", "--logfile=/tmp/nxd_model/encoding/_tp0_bk0/log-neuron-cc.txt"]
neuronxcc-2.21.18209.0+043b1bf7/MODULE_f9cc02790e4e79aee5af+fb4cc044/model.done ADDED
File without changes
neuronxcc-2.21.18209.0+043b1bf7/MODULE_f9cc02790e4e79aee5af+fb4cc044/model.hlo_module.pb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e6bc7e58b8a4764b25b574056cafa952288e53e61f35ed4dab2f593c81fd506
3
+ size 490318
neuronxcc-2.21.18209.0+043b1bf7/MODULE_f9cc02790e4e79aee5af+fb4cc044/model.neff ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5680c7be16ac4998a9dad2480d56b28005084ecbb9ff31878565ec7ba8057d13
3
+ size 19364864