Upload config.json with huggingface_hub
Browse files- config.json +28 -0
config.json
ADDED
|
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"d_model": 384,
|
| 3 |
+
"n_layers": 8,
|
| 4 |
+
"n_heads": 8,
|
| 5 |
+
"n_kv_heads": 4,
|
| 6 |
+
"d_ff": 1024,
|
| 7 |
+
"K": 16,
|
| 8 |
+
"max_len": 512,
|
| 9 |
+
"entropy_reg": 0.02,
|
| 10 |
+
"batch_size": 16,
|
| 11 |
+
"seq_len": 256,
|
| 12 |
+
"lr": 0.0003,
|
| 13 |
+
"steps": 30000,
|
| 14 |
+
"warmup_steps": 500,
|
| 15 |
+
"val_every": 500,
|
| 16 |
+
"save_every": 2500,
|
| 17 |
+
"probe_steps": [
|
| 18 |
+
1500,
|
| 19 |
+
5000,
|
| 20 |
+
15000,
|
| 21 |
+
30000
|
| 22 |
+
],
|
| 23 |
+
"vocab_size": 50257,
|
| 24 |
+
"n_params": 37107592,
|
| 25 |
+
"architecture": "CDM_V2_code",
|
| 26 |
+
"device": "cuda",
|
| 27 |
+
"dataset": "bigcode/starcoderdata:python"
|
| 28 |
+
}
|