hdallatorre
commited on
Commit
•
ebe8e9e
1
Parent(s):
bedc239
feat: Put jax checkpoint in a dedicated folder
Browse files
jax_model/hyperparams.json
ADDED
@@ -0,0 +1,26 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{"alphabet_size": 4107,
|
2 |
+
"pad_token_id": 1,
|
3 |
+
"mask_token_id": 2,
|
4 |
+
"max_positions": 2048,
|
5 |
+
"k_for_kmers": 6,
|
6 |
+
"embed_scale": 1.0,
|
7 |
+
"emb_layer_norm_before": false,
|
8 |
+
"attention_heads": 16,
|
9 |
+
"key_size": 32,
|
10 |
+
"embed_dim": 512,
|
11 |
+
"ffn_embed_dim": 2048,
|
12 |
+
"num_layers": 12,
|
13 |
+
"positional_embedding": null,
|
14 |
+
"lm_head": "roberta",
|
15 |
+
"add_bias_kv": false,
|
16 |
+
"add_bias_ffn": false,
|
17 |
+
"use_rotary_embedding": true,
|
18 |
+
"ffn_activation_name": "swish",
|
19 |
+
"use_glu_in_ffn": true,
|
20 |
+
"mask_before_attention": false,
|
21 |
+
"token_dropout": false,
|
22 |
+
"masking_ratio": 0.0,
|
23 |
+
"masking_prob": 0.0,
|
24 |
+
"use_gradient_checkpointing": false,
|
25 |
+
"embeddings_layers_to_save": [],
|
26 |
+
"attention_maps_to_save": []}
|
pytree_ckpt.joblib → jax_model/pytree_ckpt.joblib
RENAMED
File without changes
|