Upload config.json with huggingface_hub
Browse files- config.json +24 -0
config.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architecture": "DiffusionLM",
|
| 3 |
+
"T": 200,
|
| 4 |
+
"d_model": 256,
|
| 5 |
+
"n_layers": 4,
|
| 6 |
+
"d_ff": 1024,
|
| 7 |
+
"n_heads": 4,
|
| 8 |
+
"a_len": 4,
|
| 9 |
+
"q_len": 28,
|
| 10 |
+
"noise_schedule": "sqrt_power_law",
|
| 11 |
+
"clamping": "euclidean",
|
| 12 |
+
"loss": "mse_ce",
|
| 13 |
+
"decoder_causal_mask": false,
|
| 14 |
+
"vocab_size": 516,
|
| 15 |
+
"params": 8024576,
|
| 16 |
+
"training_checkpoint": "Phase 3 Reconstructed",
|
| 17 |
+
"training_steps": 500,
|
| 18 |
+
"training_lr": 0.0003,
|
| 19 |
+
"training_seed": 42,
|
| 20 |
+
"dataset": "prajwaldongre/riddles-a-synthetic-riddle-dataset-for-nlp",
|
| 21 |
+
"train_exact_match": 87.5,
|
| 22 |
+
"heldout_exact_match": 47.0,
|
| 23 |
+
"heldout_k10_exact_match": 47.0
|
| 24 |
+
}
|