Upload metrics.jsonl with huggingface_hub
Browse files- metrics.jsonl +1 -0
metrics.jsonl
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"timestamp": 1787808525.500145, "elapsed_hours": 0.0018039745754665798, "event": "start", "config": {"vocab_size": 50304, "d_model": 896, "n_layers": 18, "n_heads": 14, "n_kv_heads": 2, "max_len": 2048, "rope_theta": 500000.0, "window_size": 512, "pattern_mult": 2.5, "reason_mult": 0.75, "reason_depth": 2, "gate_hidden": 64, "gate_init": 0.0, "diff_lambda_init": 0.8, "tie_embeddings": true, "sft_lr": 3e-05, "sft_min_lr": 3e-06, "warmup": 250, "epochs": 60, "micro_batch": 2, "grad_accum": 2, "grad_clip": 1.0, "wd": 0.01, "log_every": 20, "sample_every": 150, "ckpt_every": 750, "max_hours": 11.0, "max_examples": 1096, "end_token_weight": 15.0, "tag_close_weight": 20.0, "tag_open_weight": 15.0, "min_thinking_chars": 80, "min_reasoning_chars": 60, "early_stop_window": 6, "early_stop_open_pct": 80.0, "early_stop_close_pct": 80.0, "early_stop_end_pct": 80.0, "early_stop_reason_pct": 70.0, "early_stop_min_samples": 12}, "n_examples": 1094}
|