agent=agent_hammer train_key=034e68fbf41b8489 (recovery)
Browse files
034e68fbf41b8489/config.json
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"act_cache_key": "63abe7e716cc7392",
|
| 3 |
+
"agent": "agent_hammer",
|
| 4 |
+
"arch": "tsae_paper",
|
| 5 |
+
"arch_version": "1.0.0",
|
| 6 |
+
"datasource": "toy_coupled_K10_M20_d256",
|
| 7 |
+
"saved_ts": "2026-05-06T22:49:33Z",
|
| 8 |
+
"seed": 2,
|
| 9 |
+
"train_key": "034e68fbf41b8489",
|
| 10 |
+
"training_cfg": {
|
| 11 |
+
"arch_hparams_override": {
|
| 12 |
+
"d_sae": 40,
|
| 13 |
+
"k_pos": 12
|
| 14 |
+
},
|
| 15 |
+
"batch_size": 1024,
|
| 16 |
+
"bricken_enabled": false,
|
| 17 |
+
"bricken_max_resample_fraction": 0.5,
|
| 18 |
+
"bricken_min_fires": 1,
|
| 19 |
+
"bricken_n_check": 2048,
|
| 20 |
+
"bricken_resample_every": 500,
|
| 21 |
+
"dead_threshold_tokens": 10000000,
|
| 22 |
+
"ema_auxk_alpha": 0.03125,
|
| 23 |
+
"learning_rate": 0.0003,
|
| 24 |
+
"n_steps": 30000,
|
| 25 |
+
"optimizer": "adam",
|
| 26 |
+
"plateau_early_stop": false,
|
| 27 |
+
"plateau_min_delta": 0.0001,
|
| 28 |
+
"plateau_patience": 5000,
|
| 29 |
+
"precision": "bf16",
|
| 30 |
+
"train_window_size": 2,
|
| 31 |
+
"warmup_steps": 1000
|
| 32 |
+
}
|
| 33 |
+
}
|
034e68fbf41b8489/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1d593d607f7a7e7f2bb45ba6dd3f1430cc80fc1b3568703ef17172f9385b5312
|
| 3 |
+
size 83992
|