Upload FULL128best (W&B 7rpsn108) config.json
Browse files
block_cdlm_FULL128best-s42-owt-sub-Lp16-sc0.5-fixedproposal-block-A1.5-cmstandard_20260910_155449/openwebtext-train_Lp16/config.json
ADDED
|
@@ -0,0 +1,36 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"attn_backend": "sdpa",
|
| 3 |
+
"attn_variant": "block",
|
| 4 |
+
"block_size": 16,
|
| 5 |
+
"clean_cond_init": "gamma_min",
|
| 6 |
+
"clean_gamma": null,
|
| 7 |
+
"clean_modulation": "standard",
|
| 8 |
+
"cond_dim": 128,
|
| 9 |
+
"decouple_time": true,
|
| 10 |
+
"density_residual": false,
|
| 11 |
+
"dropout": 0.1,
|
| 12 |
+
"embed_grad_path": "both",
|
| 13 |
+
"embedding_norm_method": "layernorm",
|
| 14 |
+
"gamma_stagger": 0.0,
|
| 15 |
+
"gumbel_cutoff": 1e-05,
|
| 16 |
+
"gumbel_entropy": -1,
|
| 17 |
+
"gumbel_loc": 4.723,
|
| 18 |
+
"gumbel_scale": 0.852,
|
| 19 |
+
"hidden_size": 768,
|
| 20 |
+
"model_length": 128,
|
| 21 |
+
"model_type": "BlockLangFlow",
|
| 22 |
+
"n_blocks": 12,
|
| 23 |
+
"n_heads": 12,
|
| 24 |
+
"prefix_cross_attention": false,
|
| 25 |
+
"prefix_gate": "off",
|
| 26 |
+
"prefix_gate_g0": 6.0,
|
| 27 |
+
"prefix_gate_max": 8.0,
|
| 28 |
+
"prefix_gate_tau": 1.0,
|
| 29 |
+
"prefix_mode": "concat",
|
| 30 |
+
"self_conditioning": true,
|
| 31 |
+
"time_conditioning": "adaln",
|
| 32 |
+
"transformers_version": "5.13.0",
|
| 33 |
+
"use_bias": true,
|
| 34 |
+
"use_normalized_embedding": true,
|
| 35 |
+
"vocab_size": 50257
|
| 36 |
+
}
|