Commit
d16d9f6
• 1
Parent(s):
fa81add
add mass mainstream weights
Browse files
mass/mainstream/base_uncased_slab_desc_from_lp.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9026b2d9f001a731c7800f9bbaf2e43675cd4ff50e3a45fbcd1cec4d29db0565
|
| 3 |
+
size 493928102
|
mass/mainstream/base_uncased_slab_desc_from_title.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:75d56f88d39ea4e7159c0d6e9b56582847e128294a36ae3d42e71ac2574dd33f
|
| 3 |
+
size 493928102
|
mass/mainstream/base_uncased_slab_title_from_lp.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c95f45ed5e8b8580a006711093023967535d6f622656b4fadcad1beacb5e2078
|
| 3 |
+
size 493928102
|
mass/mainstream/mass-base-uncased-config.json
ADDED
|
@@ -0,0 +1,37 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"encoder_embed_dim": 768,
|
| 3 |
+
"encoder_ffn_embed_dim": 3072,
|
| 4 |
+
"encoder_attention_heads": 12,
|
| 5 |
+
"encoder_layers": 6,
|
| 6 |
+
"encoder_normalize_before": 0,
|
| 7 |
+
"encoder_learned_pos": 1,
|
| 8 |
+
"decoder_learned_pos": 1,
|
| 9 |
+
"decoder_embed_dim": 768,
|
| 10 |
+
"decoder_ffn_embed_dim": 3072,
|
| 11 |
+
"decoder_layers": 6,
|
| 12 |
+
"decoder_attention_heads": 12,
|
| 13 |
+
"decoder_normalize_before": 0,
|
| 14 |
+
"attention_dropout": 0.1,
|
| 15 |
+
"activation_dropout": 0.1,
|
| 16 |
+
"activation_fn": "gelu",
|
| 17 |
+
"dropout": 0.1,
|
| 18 |
+
"adaptive_softmax_dropout": 0,
|
| 19 |
+
"share_decoder_input_output_embed": 1,
|
| 20 |
+
"share_all_embeddings": 1,
|
| 21 |
+
"no_token_positional_embeddings": 0,
|
| 22 |
+
"no_cross_attention": 0,
|
| 23 |
+
"cross_self_attention": 0,
|
| 24 |
+
"decoder_output_dim": 768,
|
| 25 |
+
"decoder_input_dim": 768,
|
| 26 |
+
"no_scale_embedding": 0,
|
| 27 |
+
"adaptive_input": 0,
|
| 28 |
+
"tie_adaptive_weights": 0,
|
| 29 |
+
"layernorm_embedding": 1,
|
| 30 |
+
"encoder_layerdrop": 0,
|
| 31 |
+
"decoder_layerdrop": 0,
|
| 32 |
+
"quant_noise_pq": 0,
|
| 33 |
+
"quant_noise_pq_block_size": 8,
|
| 34 |
+
"quant_noise_scalar": 0,
|
| 35 |
+
"is_encoder_decoder": 1,
|
| 36 |
+
"vocab_size": 30522
|
| 37 |
+
}
|
mass/mainstream/mass-base-uncased-vocab.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|