Upload metrics.jsonl with huggingface_hub

metrics.jsonl

@@ -1 +1 @@
-{"timestamp": 1787806902.5377326, "elapsed_hours": 0.005231806702084012, "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": 20, "epochs": 3, "micro_batch": 2, "grad_accum": 2, "grad_clip": 1.0, "wd": 0.01, "log_every": 10, "sample_every": 75, "ckpt_every": 300, "max_hours": 11.0, "max_examples": 1096, "end_token_weight": 8.0, "tag_close_weight": 10.0, "tag_open_weight": 8.0}, "n_examples": 1094}
+{"timestamp": 1787807647.8245602, "elapsed_hours": 0.004096345106760661, "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": 20, "epochs": 3, "micro_batch": 2, "grad_accum": 2, "grad_clip": 1.0, "wd": 0.01, "log_every": 10, "sample_every": 75, "ckpt_every": 300, "max_hours": 11.0, "max_examples": 1096, "end_token_weight": 50.0, "tag_close_weight": 60.0, "tag_open_weight": 50.0}, "n_examples": 1094}