Upload folder using huggingface_hub
Browse files- transformer/config.json +46 -0
- transformer/model.safetensors +3 -0
transformer/config.json
ADDED
|
@@ -0,0 +1,46 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model": {
|
| 3 |
+
"name": "transformer",
|
| 4 |
+
"d_model": 256,
|
| 5 |
+
"n_layers": 4,
|
| 6 |
+
"n_heads": 8,
|
| 7 |
+
"d_ff": 1024,
|
| 8 |
+
"dropout": 0.0,
|
| 9 |
+
"causal": true,
|
| 10 |
+
"context": 1024,
|
| 11 |
+
"overlap": 256,
|
| 12 |
+
"grad_ckpt": false
|
| 13 |
+
},
|
| 14 |
+
"train": {
|
| 15 |
+
"steps": 10000,
|
| 16 |
+
"batch_size": 16,
|
| 17 |
+
"seq_len": 1024,
|
| 18 |
+
"lr": 0.001,
|
| 19 |
+
"min_lr_ratio": 0.1,
|
| 20 |
+
"warmup": 1000,
|
| 21 |
+
"weight_decay": 0.1,
|
| 22 |
+
"grad_clip": 1.0,
|
| 23 |
+
"log_every": 50,
|
| 24 |
+
"eval_every": 1000,
|
| 25 |
+
"eval_batch_tokens": 16384,
|
| 26 |
+
"seed": 0
|
| 27 |
+
},
|
| 28 |
+
"params": 3184156,
|
| 29 |
+
"data": {
|
| 30 |
+
"pairs": 569670,
|
| 31 |
+
"dropped": 0,
|
| 32 |
+
"null_symbols": "!'+,-./:;<=>?[]_{}"
|
| 33 |
+
},
|
| 34 |
+
"best": {
|
| 35 |
+
"char_accuracy": 0.16903901609255426,
|
| 36 |
+
"word_accuracy": 0.0,
|
| 37 |
+
"precision": 0.010207666950340519,
|
| 38 |
+
"recall": 0.03713926543117077,
|
| 39 |
+
"f1": 0.010970885604912292,
|
| 40 |
+
"levenshtein": 750.7632263660017,
|
| 41 |
+
"normalized_levenshtein": 0.8309598616906798,
|
| 42 |
+
"samples": 1153,
|
| 43 |
+
"loss": 2.528406199725198,
|
| 44 |
+
"step": 2000
|
| 45 |
+
}
|
| 46 |
+
}
|
transformer/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5e06a2894c0c30d0ae81ee360c1830446c41a1010aad43ac6ee41ca807e06caa
|
| 3 |
+
size 12740576
|