Elias Frantar
commited on
Commit
·
21707b7
1
Parent(s):
7515601
Added files
Browse files
config.json
ADDED
|
@@ -0,0 +1,42 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_router_probs": false,
|
| 3 |
+
"architectures": [
|
| 4 |
+
"SwitchTransformersForConditionalGeneration"
|
| 5 |
+
],
|
| 6 |
+
"d_ff": 3072,
|
| 7 |
+
"d_kv": 64,
|
| 8 |
+
"d_model": 768,
|
| 9 |
+
"decoder_sparse_step": 2,
|
| 10 |
+
"dense_act_fn": "relu",
|
| 11 |
+
"dropout_rate": 0.1,
|
| 12 |
+
"encoder_sparse_step": 2,
|
| 13 |
+
"eos_token_id": 1,
|
| 14 |
+
"expert_capacity": 1024,
|
| 15 |
+
"feed_forward_proj": "relu",
|
| 16 |
+
"initializer_factor": 1.0,
|
| 17 |
+
"is_encoder_decoder": true,
|
| 18 |
+
"is_gated_act": false,
|
| 19 |
+
"layer_norm_epsilon": 1e-06,
|
| 20 |
+
"model_type": "switch_transformers",
|
| 21 |
+
"num_decoder_layers": 12,
|
| 22 |
+
"num_experts": 128,
|
| 23 |
+
"num_heads": 12,
|
| 24 |
+
"num_layers": 12,
|
| 25 |
+
"num_selected_experts": 1,
|
| 26 |
+
"num_sparse_decoder_layers": 6,
|
| 27 |
+
"num_sparse_encoder_layers": 6,
|
| 28 |
+
"pad_token_id": 0,
|
| 29 |
+
"relative_attention_max_distance": 128,
|
| 30 |
+
"relative_attention_num_buckets": 32,
|
| 31 |
+
"router_aux_loss_coef": 0.001,
|
| 32 |
+
"router_bias": false,
|
| 33 |
+
"router_dtype": "float32",
|
| 34 |
+
"router_ignore_padding_tokens": false,
|
| 35 |
+
"router_jitter_noise": 0,
|
| 36 |
+
"router_type": "tokens_masked",
|
| 37 |
+
"router_z_loss_coef": 0.001,
|
| 38 |
+
"torch_dtype": "bfloat16",
|
| 39 |
+
"transformers_version": "4.28.0.dev0",
|
| 40 |
+
"use_cache": false,
|
| 41 |
+
"vocab_size": 32128
|
| 42 |
+
}
|
exp00.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:91728cd98cff8d63883cde433f9f685ab0e1207fcba449462a2b7ad9c5c4b2ee
|
| 3 |
+
size 73052053
|
exp01.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9fb04bb4ff47181db747617a4e0af30be9f055eef1e85cd49a34c8291b6e7b70
|
| 3 |
+
size 68368597
|
exp02.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cca8ca2807443b8420af671770ffccfe486faa1f7ef9bd27568253e4c0402b3a
|
| 3 |
+
size 68930645
|
exp03.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3d16a974f43cf349ed9fd225330e700a67fb73b4b3e9dc00fa1df411fa31a3ca
|
| 3 |
+
size 68419029
|
exp04.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:227cfceead58d97aad489192cd5d153614e3d142e4d87cd2b87bd0273948d4cf
|
| 3 |
+
size 66393813
|
exp05.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cbe50d16156b6756e5c6b24ede3f2da8298972f805d66fef5c11fd74814872f6
|
| 3 |
+
size 68185749
|
exp06.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cb42af7063f2de6a47b303e3d92ebe2dc2dfddb6c2fcac24d4a739fa81117e2b
|
| 3 |
+
size 74270037
|
exp07.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8e0248c6be91a0cad8d9cced66a65c7d037a35fa09202cf244a0654f90414711
|
| 3 |
+
size 68103125
|
exp08.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:31a18f2665fec4d783ece0d601e98361971698223d15654a8c7d44f44b5aeaa5
|
| 3 |
+
size 69254549
|
exp09.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3bccb9fc1bd9107a27f9f686c86e519df2a8d1375831ca8513306708339dd308
|
| 3 |
+
size 69035349
|
exp10.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:181e729902f0f9b0074796a191b0a0c2751111c488676382355501c8ac9fd6ae
|
| 3 |
+
size 69697813
|
exp11.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a5c345192b0d9089fbdf7c86b7b8cb748a9be5642db750bbc28570cb2c1acd98
|
| 3 |
+
size 69390357
|
noexp.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2b82c78a9527ce441da27b36e57b338142a739facde6e6de088a12130b4d403b
|
| 3 |
+
size 434367909
|
sizes.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:66dacfe9ed9d9ca3d6a3ecefee62e9ec382f64b1d9f54b1f3f192775d765d493
|
| 3 |
+
size 194091
|