Hugging Face's logo

aws-neuron
/
optimum-neuron-cache

Model card Files
xet
Community
648
dacorvo HF Staff commited on
Commit
7499a65
·
verified ·
1 Parent(s): 26f6082

Synchronizing local compiler cache.

Browse files
Files changed (8) hide show
  1. neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.0/granite/ibm-granite/granite-3.1-2b-instruct/43a40fa53d181d38cbec.json +58 -0
  2. neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.0/llama/meta-llama/Meta-Llama-3.1-8B/9aaca0d9f5b6c41a6516.json +62 -0
  3. neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.0/llama/unsloth/Llama-3.2-1B-Instruct/c9c073b06914f52c8ee3.json +63 -0
  4. neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.0/phi3/microsoft/Phi-3-mini-4k-instruct/ecc9e5085068b585af4a.json +62 -0
  5. neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.0/qwen2/Qwen/Qwen2.5-0.5B/5b64bd0534032301f2e5.json +82 -0
  6. neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.0/qwen2/Qwen/Qwen2.5-0.5B/9d0b9126b8fed3c361f7.json +82 -0
  7. neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.0/qwen3/Qwen/Qwen3-0.6B/837516bcf385ea54670f.json +87 -0
  8. neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.0/smollm3/HuggingFaceTB/SmolLM3-3B/00cee8c1b2bf2f6555e0.json +134 -0
neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.0/granite/ibm-granite/granite-3.1-2b-instruct/43a40fa53d181d38cbec.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
+ "embedding_multiplier": 12.0,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 2048,
14
+ "initializer_range": 0.02,
15
+ "intermediate_size": 8192,
16
+ "logits_scaling": 8.0,
17
+ "max_position_embeddings": 131072,
18
+ "mlp_bias": false,
19
+ "model_type": "granite",
20
+ "neuron": {
21
+ "_serialized_key": "NxDNeuronConfig",
22
+ "batch_size": 4,
23
+ "capacity_factor": null,
24
+ "checkpoint_id": "ibm-granite/granite-3.1-2b-instruct",
25
+ "checkpoint_revision": "bbc2aed595bd38bd770263dc3ab831db9794441d",
26
+ "continuous_batching": true,
27
+ "enable_bucketing": false,
28
+ "ep_degree": 1,
29
+ "fused_qkv": true,
30
+ "glu_mlp": true,
31
+ "local_ranks_size": 2,
32
+ "max_batch_size": 4,
33
+ "max_context_length": 4096,
34
+ "max_topk": 256,
35
+ "n_active_tokens": 4096,
36
+ "neuronxcc_version": "2.21.18209.0+043b1bf7",
37
+ "on_device_sampling": true,
38
+ "optimum_neuron_version": "0.4.0",
39
+ "output_logits": false,
40
+ "pp_degree": 1,
41
+ "sequence_length": 4096,
42
+ "speculation_length": 0,
43
+ "start_rank_id": 0,
44
+ "target": "trn1",
45
+ "torch_dtype": "bfloat16",
46
+ "tp_degree": 2
47
+ },
48
+ "num_attention_heads": 32,
49
+ "num_hidden_layers": 40,
50
+ "num_key_value_heads": 8,
51
+ "residual_multiplier": 0.22,
52
+ "rms_norm_eps": 1e-05,
53
+ "rope_scaling": null,
54
+ "rope_theta": 5000000.0,
55
+ "tie_word_embeddings": true,
56
+ "use_cache": true,
57
+ "vocab_size": 49155
58
+ }
neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.0/llama/meta-llama/Meta-Llama-3.1-8B/9aaca0d9f5b6c41a6516.json ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "meta-llama/Meta-Llama-3.1-8B",
4
+ "_task": "text-generation",
5
+ "architectures": [
6
+ "LlamaForCausalLM"
7
+ ],
8
+ "attention_bias": false,
9
+ "attention_dropout": 0.0,
10
+ "head_dim": 128,
11
+ "hidden_act": "silu",
12
+ "hidden_size": 4096,
13
+ "initializer_range": 0.02,
14
+ "intermediate_size": 14336,
15
+ "max_position_embeddings": 131072,
16
+ "mlp_bias": false,
17
+ "model_type": "llama",
18
+ "neuron": {
19
+ "_serialized_key": "NxDNeuronConfig",
20
+ "batch_size": 1,
21
+ "capacity_factor": null,
22
+ "checkpoint_id": "meta-llama/Meta-Llama-3.1-8B",
23
+ "checkpoint_revision": "d04e592bb4f6aa9cfee91e2e20afa771667e1d4b",
24
+ "continuous_batching": false,
25
+ "enable_bucketing": false,
26
+ "ep_degree": 1,
27
+ "fused_qkv": true,
28
+ "glu_mlp": true,
29
+ "local_ranks_size": 2,
30
+ "max_batch_size": 1,
31
+ "max_context_length": 4096,
32
+ "max_topk": 256,
33
+ "n_active_tokens": 4096,
34
+ "neuronxcc_version": "2.21.18209.0+043b1bf7",
35
+ "on_device_sampling": true,
36
+ "optimum_neuron_version": "0.4.0",
37
+ "output_logits": false,
38
+ "pp_degree": 1,
39
+ "sequence_length": 4096,
40
+ "speculation_length": 0,
41
+ "start_rank_id": 0,
42
+ "target": "trn1",
43
+ "torch_dtype": "bfloat16",
44
+ "tp_degree": 2
45
+ },
46
+ "num_attention_heads": 32,
47
+ "num_hidden_layers": 32,
48
+ "num_key_value_heads": 8,
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/0.4.0/llama/unsloth/Llama-3.2-1B-Instruct/c9c073b06914f52c8ee3.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
+ "head_dim": 64,
11
+ "hidden_act": "silu",
12
+ "hidden_size": 2048,
13
+ "initializer_range": 0.02,
14
+ "intermediate_size": 8192,
15
+ "max_position_embeddings": 131072,
16
+ "mlp_bias": false,
17
+ "model_type": "llama",
18
+ "neuron": {
19
+ "_serialized_key": "NxDNeuronConfig",
20
+ "batch_size": 4,
21
+ "capacity_factor": null,
22
+ "checkpoint_id": "unsloth/Llama-3.2-1B-Instruct",
23
+ "checkpoint_revision": "5a8abab4a5d6f164389b1079fb721cfab8d7126c",
24
+ "continuous_batching": true,
25
+ "enable_bucketing": false,
26
+ "ep_degree": 1,
27
+ "fused_qkv": true,
28
+ "glu_mlp": true,
29
+ "local_ranks_size": 2,
30
+ "max_batch_size": 4,
31
+ "max_context_length": 4096,
32
+ "max_topk": 256,
33
+ "n_active_tokens": 4096,
34
+ "neuronxcc_version": "2.21.18209.0+043b1bf7",
35
+ "on_device_sampling": true,
36
+ "optimum_neuron_version": "0.4.0",
37
+ "output_logits": false,
38
+ "pp_degree": 1,
39
+ "sequence_length": 4096,
40
+ "speculation_length": 0,
41
+ "start_rank_id": 0,
42
+ "target": "trn1",
43
+ "torch_dtype": "float16",
44
+ "tp_degree": 2
45
+ },
46
+ "num_attention_heads": 32,
47
+ "num_hidden_layers": 16,
48
+ "num_key_value_heads": 8,
49
+ "pretraining_tp": 1,
50
+ "rms_norm_eps": 1e-05,
51
+ "rope_scaling": {
52
+ "factor": 32.0,
53
+ "high_freq_factor": 4.0,
54
+ "low_freq_factor": 1.0,
55
+ "original_max_position_embeddings": 8192,
56
+ "rope_type": "llama3"
57
+ },
58
+ "rope_theta": 500000.0,
59
+ "tie_word_embeddings": true,
60
+ "unsloth_fixed": true,
61
+ "use_cache": true,
62
+ "vocab_size": 128256
63
+ }
neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.0/phi3/microsoft/Phi-3-mini-4k-instruct/ecc9e5085068b585af4a.json ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_entry_class": "SingleModelCacheEntry",
3
+ "_model_id": "microsoft/Phi-3-mini-4k-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
+ "embd_pdrop": 0.0,
15
+ "hidden_act": "silu",
16
+ "hidden_size": 3072,
17
+ "initializer_range": 0.02,
18
+ "intermediate_size": 8192,
19
+ "max_position_embeddings": 4096,
20
+ "model_type": "phi3",
21
+ "neuron": {
22
+ "_serialized_key": "NxDNeuronConfig",
23
+ "batch_size": 4,
24
+ "capacity_factor": null,
25
+ "checkpoint_id": "microsoft/Phi-3-mini-4k-instruct",
26
+ "checkpoint_revision": "0a67737cc96d2554230f90338b163bc6380a2a85",
27
+ "continuous_batching": true,
28
+ "enable_bucketing": false,
29
+ "ep_degree": 1,
30
+ "fused_qkv": true,
31
+ "glu_mlp": true,
32
+ "local_ranks_size": 2,
33
+ "max_batch_size": 4,
34
+ "max_context_length": 4096,
35
+ "max_topk": 256,
36
+ "n_active_tokens": 4096,
37
+ "neuronxcc_version": "2.21.18209.0+043b1bf7",
38
+ "on_device_sampling": true,
39
+ "optimum_neuron_version": "0.4.0",
40
+ "output_logits": false,
41
+ "pp_degree": 1,
42
+ "sequence_length": 4096,
43
+ "speculation_length": 0,
44
+ "start_rank_id": 0,
45
+ "target": "trn1",
46
+ "torch_dtype": "bfloat16",
47
+ "tp_degree": 2
48
+ },
49
+ "num_attention_heads": 32,
50
+ "num_hidden_layers": 32,
51
+ "num_key_value_heads": 32,
52
+ "original_max_position_embeddings": 4096,
53
+ "partial_rotary_factor": 1.0,
54
+ "resid_pdrop": 0.0,
55
+ "rms_norm_eps": 1e-05,
56
+ "rope_scaling": null,
57
+ "rope_theta": 10000.0,
58
+ "sliding_window": 2047,
59
+ "tie_word_embeddings": false,
60
+ "use_cache": true,
61
+ "vocab_size": 32064
62
+ }
neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.0/qwen2/Qwen/Qwen2.5-0.5B/5b64bd0534032301f2e5.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
+ "hidden_act": "silu",
10
+ "hidden_size": 896,
11
+ "initializer_range": 0.02,
12
+ "intermediate_size": 4864,
13
+ "layer_types": [
14
+ "full_attention",
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
+ ],
39
+ "max_position_embeddings": 32768,
40
+ "max_window_layers": 24,
41
+ "model_type": "qwen2",
42
+ "neuron": {
43
+ "_serialized_key": "NxDNeuronConfig",
44
+ "batch_size": 4,
45
+ "capacity_factor": null,
46
+ "checkpoint_id": "Qwen/Qwen2.5-0.5B",
47
+ "checkpoint_revision": "060db6499f32faf8b98477b0a26969ef7d8b9987",
48
+ "continuous_batching": true,
49
+ "enable_bucketing": false,
50
+ "ep_degree": 1,
51
+ "fused_qkv": false,
52
+ "glu_mlp": true,
53
+ "local_ranks_size": 2,
54
+ "max_batch_size": 4,
55
+ "max_context_length": 4096,
56
+ "max_topk": 256,
57
+ "n_active_tokens": 4096,
58
+ "neuronxcc_version": "2.21.18209.0+043b1bf7",
59
+ "on_device_sampling": false,
60
+ "optimum_neuron_version": "0.4.0",
61
+ "output_logits": false,
62
+ "pp_degree": 1,
63
+ "sequence_length": 4096,
64
+ "speculation_length": 0,
65
+ "start_rank_id": 0,
66
+ "target": "trn1",
67
+ "torch_dtype": "float16",
68
+ "tp_degree": 2
69
+ },
70
+ "num_attention_heads": 14,
71
+ "num_hidden_layers": 24,
72
+ "num_key_value_heads": 2,
73
+ "rms_norm_eps": 1e-06,
74
+ "rope_scaling": null,
75
+ "rope_theta": 1000000.0,
76
+ "sliding_window": null,
77
+ "tie_word_embeddings": true,
78
+ "use_cache": true,
79
+ "use_mrope": false,
80
+ "use_sliding_window": false,
81
+ "vocab_size": 151936
82
+ }
neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.0/qwen2/Qwen/Qwen2.5-0.5B/9d0b9126b8fed3c361f7.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
+ "hidden_act": "silu",
10
+ "hidden_size": 896,
11
+ "initializer_range": 0.02,
12
+ "intermediate_size": 4864,
13
+ "layer_types": [
14
+ "full_attention",
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
+ ],
39
+ "max_position_embeddings": 32768,
40
+ "max_window_layers": 24,
41
+ "model_type": "qwen2",
42
+ "neuron": {
43
+ "_serialized_key": "NxDNeuronConfig",
44
+ "batch_size": 1,
45
+ "capacity_factor": null,
46
+ "checkpoint_id": "Qwen/Qwen2.5-0.5B",
47
+ "checkpoint_revision": "060db6499f32faf8b98477b0a26969ef7d8b9987",
48
+ "continuous_batching": false,
49
+ "enable_bucketing": false,
50
+ "ep_degree": 1,
51
+ "fused_qkv": false,
52
+ "glu_mlp": true,
53
+ "local_ranks_size": 2,
54
+ "max_batch_size": 1,
55
+ "max_context_length": 4096,
56
+ "max_topk": 256,
57
+ "n_active_tokens": 4096,
58
+ "neuronxcc_version": "2.21.18209.0+043b1bf7",
59
+ "on_device_sampling": true,
60
+ "optimum_neuron_version": "0.4.0",
61
+ "output_logits": false,
62
+ "pp_degree": 1,
63
+ "sequence_length": 4096,
64
+ "speculation_length": 0,
65
+ "start_rank_id": 0,
66
+ "target": "trn1",
67
+ "torch_dtype": "bfloat16",
68
+ "tp_degree": 2
69
+ },
70
+ "num_attention_heads": 14,
71
+ "num_hidden_layers": 24,
72
+ "num_key_value_heads": 2,
73
+ "rms_norm_eps": 1e-06,
74
+ "rope_scaling": null,
75
+ "rope_theta": 1000000.0,
76
+ "sliding_window": null,
77
+ "tie_word_embeddings": true,
78
+ "use_cache": true,
79
+ "use_mrope": false,
80
+ "use_sliding_window": false,
81
+ "vocab_size": 151936
82
+ }
neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.0/qwen3/Qwen/Qwen3-0.6B/837516bcf385ea54670f.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
+ "head_dim": 128,
11
+ "hidden_act": "silu",
12
+ "hidden_size": 1024,
13
+ "initializer_range": 0.02,
14
+ "intermediate_size": 3072,
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
+ ],
45
+ "max_position_embeddings": 40960,
46
+ "max_window_layers": 28,
47
+ "model_type": "qwen3",
48
+ "neuron": {
49
+ "_serialized_key": "NxDNeuronConfig",
50
+ "batch_size": 4,
51
+ "capacity_factor": null,
52
+ "checkpoint_id": "Qwen/Qwen3-0.6B",
53
+ "checkpoint_revision": "c1899de289a04d12100db370d81485cdf75e47ca",
54
+ "continuous_batching": true,
55
+ "enable_bucketing": 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": 4096,
62
+ "max_topk": 256,
63
+ "n_active_tokens": 4096,
64
+ "neuronxcc_version": "2.21.18209.0+043b1bf7",
65
+ "on_device_sampling": false,
66
+ "optimum_neuron_version": "0.4.0",
67
+ "output_logits": false,
68
+ "pp_degree": 1,
69
+ "sequence_length": 4096,
70
+ "speculation_length": 0,
71
+ "start_rank_id": 0,
72
+ "target": "trn1",
73
+ "torch_dtype": "bfloat16",
74
+ "tp_degree": 2
75
+ },
76
+ "num_attention_heads": 16,
77
+ "num_hidden_layers": 28,
78
+ "num_key_value_heads": 8,
79
+ "rms_norm_eps": 1e-06,
80
+ "rope_scaling": null,
81
+ "rope_theta": 1000000,
82
+ "sliding_window": null,
83
+ "tie_word_embeddings": true,
84
+ "use_cache": true,
85
+ "use_sliding_window": false,
86
+ "vocab_size": 151936
87
+ }
neuronxcc-2.21.18209.0+043b1bf7/0_REGISTRY/0.4.0/smollm3/HuggingFaceTB/SmolLM3-3B/00cee8c1b2bf2f6555e0.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
+ "hidden_act": "silu",
11
+ "hidden_size": 2048,
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 11008,
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
+ "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
+ ],
52
+ "max_position_embeddings": 65536,
53
+ "max_window_layers": 28,
54
+ "mlp_bias": false,
55
+ "model_type": "smollm3",
56
+ "neuron": {
57
+ "_serialized_key": "NxDNeuronConfig",
58
+ "batch_size": 4,
59
+ "capacity_factor": null,
60
+ "checkpoint_id": "HuggingFaceTB/SmolLM3-3B",
61
+ "checkpoint_revision": "a07cc9a04f16550a088caea529712d1d335b0ac1",
62
+ "continuous_batching": true,
63
+ "enable_bucketing": false,
64
+ "ep_degree": 1,
65
+ "fused_qkv": true,
66
+ "glu_mlp": true,
67
+ "local_ranks_size": 2,
68
+ "max_batch_size": 4,
69
+ "max_context_length": 4096,
70
+ "max_topk": 256,
71
+ "n_active_tokens": 4096,
72
+ "neuronxcc_version": "2.21.18209.0+043b1bf7",
73
+ "on_device_sampling": true,
74
+ "optimum_neuron_version": "0.4.0",
75
+ "output_logits": false,
76
+ "pp_degree": 1,
77
+ "sequence_length": 4096,
78
+ "speculation_length": 0,
79
+ "start_rank_id": 0,
80
+ "target": "trn1",
81
+ "torch_dtype": "bfloat16",
82
+ "tp_degree": 2
83
+ },
84
+ "no_rope_layer_interval": 4,
85
+ "no_rope_layers": [
86
+ 1,
87
+ 1,
88
+ 1,
89
+ 0,
90
+ 1,
91
+ 1,
92
+ 1,
93
+ 0,
94
+ 1,
95
+ 1,
96
+ 1,
97
+ 0,
98
+ 1,
99
+ 1,
100
+ 1,
101
+ 0,
102
+ 1,
103
+ 1,
104
+ 1,
105
+ 0,
106
+ 1,
107
+ 1,
108
+ 1,
109
+ 0,
110
+ 1,
111
+ 1,
112
+ 1,
113
+ 0,
114
+ 1,
115
+ 1,
116
+ 1,
117
+ 0,
118
+ 1,
119
+ 1,
120
+ 1,
121
+ 0
122
+ ],
123
+ "num_attention_heads": 16,
124
+ "num_hidden_layers": 36,
125
+ "num_key_value_heads": 4,
126
+ "pretraining_tp": 2,
127
+ "rms_norm_eps": 1e-06,
128
+ "rope_scaling": null,
129
+ "rope_theta": 5000000.0,
130
+ "sliding_window": null,
131
+ "use_cache": false,
132
+ "use_sliding_window": false,
133
+ "vocab_size": 128256
134
+ }