Hugging Face's logo

nvidia
/
OpenMath-Mistral-7B-v0.1

NeMo
English
nvidia
code
math
Model card Files Community
igitman commited on
Commit
1c3424f
•
1 Parent(s): 2b7db26

Add model files

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 +1 -0
  2. nemo_model/9a4a326142114421a1c520c7806364f6_tokenizer.model +3 -0
  3. nemo_model/9eaae26030e74c04bdd22106547a4fa4_tokenizer.model +3 -0
  4. nemo_model/model_config.yaml +121 -0
  5. nemo_model/model_weights/common.pt +3 -0
  6. nemo_model/model_weights/metadata.json +3 -0
  7. nemo_model/model_weights/model.decoder.final_layernorm.weight/.zarray +3 -0
  8. nemo_model/model_weights/model.decoder.final_layernorm.weight/0 +3 -0
  9. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_0_32.pt +3 -0
  10. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_10_32.pt +3 -0
  11. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_11_32.pt +3 -0
  12. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_12_32.pt +3 -0
  13. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_13_32.pt +3 -0
  14. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_14_32.pt +3 -0
  15. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_15_32.pt +3 -0
  16. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_16_32.pt +3 -0
  17. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_17_32.pt +3 -0
  18. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_18_32.pt +3 -0
  19. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_19_32.pt +3 -0
  20. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_1_32.pt +3 -0
  21. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_20_32.pt +3 -0
  22. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_21_32.pt +3 -0
  23. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_22_32.pt +3 -0
  24. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_23_32.pt +3 -0
  25. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_24_32.pt +3 -0
  26. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_25_32.pt +3 -0
  27. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_26_32.pt +3 -0
  28. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_27_32.pt +3 -0
  29. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_28_32.pt +3 -0
  30. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_29_32.pt +3 -0
  31. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_2_32.pt +3 -0
  32. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_30_32.pt +3 -0
  33. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_31_32.pt +3 -0
  34. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_3_32.pt +3 -0
  35. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_4_32.pt +3 -0
  36. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_5_32.pt +3 -0
  37. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_6_32.pt +3 -0
  38. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_7_32.pt +3 -0
  39. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_8_32.pt +3 -0
  40. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_9_32.pt +3 -0
  41. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1.layer_norm_weight/.zarray +3 -0
  42. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1.layer_norm_weight/0.0 +3 -0
  43. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1.layer_norm_weight/1.0 +3 -0
  44. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1.layer_norm_weight/10.0 +3 -0
  45. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1.layer_norm_weight/11.0 +3 -0
  46. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1.layer_norm_weight/12.0 +3 -0
  47. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1.layer_norm_weight/13.0 +3 -0
  48. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1.layer_norm_weight/14.0 +3 -0
  49. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1.layer_norm_weight/15.0 +3 -0
  50. nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1.layer_norm_weight/16.0 +3 -0
.gitattributes CHANGED
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ nemo_model/model_weights/** filter=lfs diff=lfs merge=lfs -text
nemo_model/9a4a326142114421a1c520c7806364f6_tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dadfd56d766715c61d2ef780a525ab43b8e6da4de6865bda3d95fdef5e134055
3
+ size 493443
nemo_model/9eaae26030e74c04bdd22106547a4fa4_tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dadfd56d766715c61d2ef780a525ab43b8e6da4de6865bda3d95fdef5e134055
3
+ size 493443
nemo_model/model_config.yaml ADDED
@@ -0,0 +1,121 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ mcore_gpt: true
2
+ micro_batch_size: 4
3
+ global_batch_size: 8
4
+ tensor_model_parallel_size: 1
5
+ pipeline_model_parallel_size: 1
6
+ virtual_pipeline_model_parallel_size: null
7
+ encoder_seq_length: 32768
8
+ max_position_embeddings: 32768
9
+ num_layers: 32
10
+ hidden_size: 4096
11
+ ffn_hidden_size: 14336
12
+ num_attention_heads: 32
13
+ init_method_std: 0.02
14
+ use_scaled_init_method: true
15
+ hidden_dropout: 0.0
16
+ attention_dropout: 0.0
17
+ ffn_dropout: 0.0
18
+ kv_channels: null
19
+ apply_query_key_layer_scaling: true
20
+ normalization: rmsnorm
21
+ layernorm_epsilon: 1.0e-05
22
+ do_layer_norm_weight_decay: false
23
+ make_vocab_size_divisible_by: 128
24
+ pre_process: true
25
+ post_process: true
26
+ persist_layer_norm: true
27
+ bias: false
28
+ activation: fast-swiglu
29
+ headscale: false
30
+ transformer_block_type: pre_ln
31
+ openai_gelu: false
32
+ normalize_attention_scores: true
33
+ position_embedding_type: rope
34
+ rotary_percentage: 1.0
35
+ attention_type: multihead
36
+ share_embeddings_and_output_weights: false
37
+ overlap_p2p_comm: false
38
+ batch_p2p_comm: true
39
+ num_query_groups: 8
40
+ tokenizer:
41
+ library: sentencepiece
42
+ type: null
43
+ model: nemo:9eaae26030e74c04bdd22106547a4fa4_tokenizer.model
44
+ vocab_file: null
45
+ merge_file: null
46
+ delimiter: null
47
+ sentencepiece_legacy: false
48
+ tokenizer_model: nemo:9a4a326142114421a1c520c7806364f6_tokenizer.model
49
+ native_amp_init_scale: 4294967296
50
+ native_amp_growth_interval: 1000
51
+ hysteresis: 2
52
+ fp32_residual_connection: false
53
+ fp16_lm_cross_entropy: false
54
+ megatron_amp_O2: false
55
+ grad_allreduce_chunk_size_mb: 125
56
+ grad_div_ar_fusion: true
57
+ gradient_accumulation_fusion: false
58
+ bias_activation_fusion: false
59
+ bias_dropout_add_fusion: false
60
+ masked_softmax_fusion: true
61
+ get_attention_mask_from_fusion: true
62
+ seed: 1234
63
+ resume_from_checkpoint: null
64
+ use_cpu_initialization: false
65
+ onnx_safe: false
66
+ apex_transformer_log_level: 30
67
+ gradient_as_bucket_view: true
68
+ sync_batch_comm: false
69
+ activations_checkpoint_granularity: null
70
+ activations_checkpoint_method: null
71
+ activations_checkpoint_num_layers: null
72
+ num_micro_batches_with_partial_activation_checkpoints: null
73
+ activations_checkpoint_layers_per_pipeline: null
74
+ sequence_parallel: false
75
+ transformer_engine: true
76
+ fp8: false
77
+ fp8_e4m3: false
78
+ fp8_hybrid: false
79
+ fp8_margin: 0
80
+ fp8_interval: 1
81
+ fp8_amax_history_len: 1
82
+ fp8_amax_compute_algo: most_recent
83
+ reduce_amax: true
84
+ use_emha: false
85
+ data:
86
+ index_mapping_dir: null
87
+ data_impl: mmap
88
+ splits_string: 900,50,50
89
+ seq_length: 32768
90
+ skip_warmup: true
91
+ num_workers: 2
92
+ dataloader_type: single
93
+ reset_position_ids: false
94
+ reset_attention_mask: false
95
+ eod_mask_loss: false
96
+ validation_drop_last: true
97
+ no_seqlen_plus_one_input_tokens: false
98
+ pad_samples_to_global_batch_size: false
99
+ shuffle_documents: true
100
+ nsys_profile:
101
+ enabled: false
102
+ start_step: 10
103
+ end_step: 10
104
+ ranks:
105
+ - 0
106
+ gen_shape: false
107
+ optim:
108
+ name: fused_adam
109
+ lr: 0.0002
110
+ weight_decay: 0.01
111
+ betas:
112
+ - 0.9
113
+ - 0.98
114
+ sched:
115
+ name: CosineAnnealing
116
+ warmup_steps: 500
117
+ constant_steps: 50000
118
+ min_lr: 2.0e-05
119
+ precision: 32
120
+ target: nemo.collections.nlp.models.language_modeling.megatron_gpt_model.MegatronGPTModel
121
+ nemo_version: 1.21.0
nemo_model/model_weights/common.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1c01ce38c2be3c35cd3a203cdc408e9430db0e61a741ece48927d7005f447db3
3
+ size 26223
nemo_model/model_weights/metadata.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f502ab47c9b1fe00615afb88074d49cf6f9b4c85f482734543160c5cb2f76af9
3
+ size 113
nemo_model/model_weights/model.decoder.final_layernorm.weight/.zarray ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:89126c79a8346f55cf3ba4ed68fee8f92509b6327c19e0f12658fcf705bba841
3
+ size 207
nemo_model/model_weights/model.decoder.final_layernorm.weight/0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1d58b57e44c575306e22577b7e544e659cc32d2e35dbbab00ba113100ada23cc
3
+ size 8192
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_0_32.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bfea882a79f2bdc6691f3abffbf42f72ea90a206a873ced4c23ba225be3a52b9
3
+ size 1836
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_10_32.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5ce2fde5075677bd348eda3e8fec4edeb648738c9160fdf341ea4bd3d502a774
3
+ size 1840
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_11_32.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7570532fbe5a7308c0eab584375a11ea5447c53c0f37baa45d0151dd0eb3f64c
3
+ size 1840
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_12_32.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5f25e501e97cff7e8de02864772880ba25c1f94f3df6c49aaa0b4dfec3844b1b
3
+ size 1840
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_13_32.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1342c024e8d734dba993c1b2703152ac728b93923959bd96668f521c614ff6f0
3
+ size 1840
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_14_32.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f87ac81f15e7ccca8037453d164f6b0a311df1c456912d5a5845243aa9a83107
3
+ size 1840
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_15_32.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2d6c29a320fc3bb7ca2177396a4d262d5bb686d6ee6695e957a5259707e910fc
3
+ size 1840
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_16_32.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3e033d5985303a093052da5d9f4dea431c518af71c277a87f5a23bf3a6be39ad
3
+ size 1840
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_17_32.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d9495d6607a34a2ee495c8ef1a9ef72a0c5276ada44355618992a56a7404df95
3
+ size 1840
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_18_32.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8598da7e410942c77870b4e5bde6b7a9d28391ac318438c409c9ad8e6dca52da
3
+ size 1840
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_19_32.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:06faf86285c4b17afd70ad266428a87834326e5cf2911a7a7f75acb12222e7bb
3
+ size 1840
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_1_32.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d595d940b7304a5515a42a13db5243328cf166c05d6b389dc38ba296f895b99c
3
+ size 1836
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_20_32.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5eccd309df2ac0d7db90c90896b13b0f798e8efc5614f96659dece819f797382
3
+ size 1840
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_21_32.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1d1063001f00061025ff3b86b9b03e2d4b9890486ac0fd4dd52128a0b613c637
3
+ size 1840
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_22_32.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9dfbd9465298388f8614145b4a10caab9c7f8d0d55abe62ea7b8abfc8c47d2e8
3
+ size 1840
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_23_32.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5bc13cb8c15337279cabdcfeefa23321afde337b7b1284f19824c43a0863cf76
3
+ size 1840
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_24_32.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4d8f7582bb9b53cf1c5fa153cef9175621b3dad029679c33aaf345766500e4d3
3
+ size 1840
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_25_32.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:02f4e6361c8a0592b80aa747c8a0fc6d1d51fcebc316a89af4a64697edf2c228
3
+ size 1840
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_26_32.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:35fc5ddb8b6a5e8173eff44a1a258649591b98faefeb4ae37df307cfde547a8a
3
+ size 1840
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_27_32.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:71e7685fc01ba14be5ea09d5faeb559f569b52461fd2c0cad1c16694f2a250b9
3
+ size 1840
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_28_32.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:95929b1933ba15aeb06cc40866fb9bc4849d315e5167854256cfbb604ef140f7
3
+ size 1840
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_29_32.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:19cc67352bf8ce0ded1ad8c89a8a7d1893620b64db012b54d1585e9983c90a8c
3
+ size 1840
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_2_32.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:486d703cc7a816327e1b18018996059a23adf9177172614e860956333dd2eef0
3
+ size 1836
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_30_32.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:182b1e32fa0c27a25cd6525d3fb451cacc43f098039497d197b8af232c17956f
3
+ size 1840
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_31_32.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:61a0a2cf37407340f38545d83578fb11ba1cc31dd32ba53cb6c2ef01a94e8cc8
3
+ size 1840
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_3_32.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e4320cf77a3d24b17bb878df3147154cec3781c7998cba60a0040af04b5fc791
3
+ size 1836
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_4_32.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1f48b20cd6ce31054bb776dc121e2d8c860805aba3290e19ae82d80372d0d4b4
3
+ size 1836
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_5_32.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c8c00d72ef6d697433831669aab19149eb06a9acd1b5dde8d9bddbffc239dd1d
3
+ size 1836
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_6_32.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:91e4978cb21be48bbb69d8455cc10f6e5e076dc85878033e24e2a2602a8bd6ad
3
+ size 1836
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_7_32.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:22b4ebcae96a9f703a9dbb76d7e0a9a4a8f1842cd879e739b92a40a11242f5f1
3
+ size 1836
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_8_32.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:99c1c7068de4f865aa72c7521437552e1bf92c309441fc1ad65d305302200d5e
3
+ size 1836
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1._extra_state/shard_9_32.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:31e95d1a0f3a9880b886738863ad1d3e2200e21959610023c0382eb3f863b29f
3
+ size 1836
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1.layer_norm_weight/.zarray ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2b7133d0268f56aeb1759633f4a94b185610be536a144bd4866f566a3b80e6cd
3
+ size 230
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1.layer_norm_weight/0.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9646defa8d9429b72e8d740ebeacfff43a208701c1f8c77d1848b3c2d05d70c7
3
+ size 8192
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1.layer_norm_weight/1.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9f5eac0b28baa75dd6ade6991d283536abe863f128a431254016aca71e670668
3
+ size 8192
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1.layer_norm_weight/10.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3694d1c3f4b85ff510262f953095516e3d8fe9081b1e438fad17b5fcbd359786
3
+ size 8192
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1.layer_norm_weight/11.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:40c5c6c498306f69016b7f45e8da37303e369707208c309efe381f464fafbb29
3
+ size 8192
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1.layer_norm_weight/12.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d9b418c9adf9903ae0f3f32e2bb1cdfa1cc5f1b21212cfd383cbc0a08bd10b39
3
+ size 8192
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1.layer_norm_weight/13.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ad3a5dfa3504c34c6cae84ba3706ccd2f52add12f75566b60958277ad83b7cfc
3
+ size 8192
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1.layer_norm_weight/14.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:14ff8a37953ab329f7342cf69ca801ec754e9ec0b5fe99f34c3b0fad71c89757
3
+ size 8192
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1.layer_norm_weight/15.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4295cb5cc8a70389ba9ca2de93cd30557ea2d53b38bdef2ffe4ec29f383a969f
3
+ size 8192
nemo_model/model_weights/model.decoder.layers.mlp.linear_fc1.layer_norm_weight/16.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6bab58289cd6a6ddc1d84d560ad330e775a576497946f457cd916eb228e89c54
3
+ size 8192