Upload metrics_session003.jsonl with huggingface_hub
Browse files- metrics_session003.jsonl +3 -0
metrics_session003.jsonl
CHANGED
|
@@ -37,3 +37,6 @@
|
|
| 37 |
{"kind": "train", "step": 300, "epoch": 0, "loss": 4.083635, "avg50": 4.162716, "ppl": 59.3609, "aux_loss": 0.081477, "muon_lr": 0.0021000000000000003, "adamw_lr": 7.5e-05, "tok_per_sec": 25384.8, "sec_per_step": 46.471, "vram_used_gb": 11.43, "vram_tot_gb": 95.0, "moe_dead": 0, "moe_slots": 3840, "t": 1787723781.5557492, "elapsed_h": 3.8844, "session": 3}
|
| 38 |
{"kind": "train", "step": 310, "epoch": 0, "loss": 3.957139, "avg50": 4.086978, "ppl": 52.3075, "aux_loss": 0.080664, "muon_lr": 0.00217, "adamw_lr": 7.75e-05, "tok_per_sec": 25454.1, "sec_per_step": 46.344, "vram_used_gb": 11.43, "vram_tot_gb": 95.0, "moe_dead": 0, "moe_slots": 3840, "t": 1787724207.0287333, "elapsed_h": 4.0026, "session": 3}
|
| 39 |
{"kind": "ckpt", "step": 319, "epoch": 0, "ok": true, "t": 1787724650.6797802, "elapsed_h": 4.1259, "session": 3}
|
|
|
|
|
|
|
|
|
|
|
|
| 37 |
{"kind": "train", "step": 300, "epoch": 0, "loss": 4.083635, "avg50": 4.162716, "ppl": 59.3609, "aux_loss": 0.081477, "muon_lr": 0.0021000000000000003, "adamw_lr": 7.5e-05, "tok_per_sec": 25384.8, "sec_per_step": 46.471, "vram_used_gb": 11.43, "vram_tot_gb": 95.0, "moe_dead": 0, "moe_slots": 3840, "t": 1787723781.5557492, "elapsed_h": 3.8844, "session": 3}
|
| 38 |
{"kind": "train", "step": 310, "epoch": 0, "loss": 3.957139, "avg50": 4.086978, "ppl": 52.3075, "aux_loss": 0.080664, "muon_lr": 0.00217, "adamw_lr": 7.75e-05, "tok_per_sec": 25454.1, "sec_per_step": 46.344, "vram_used_gb": 11.43, "vram_tot_gb": 95.0, "moe_dead": 0, "moe_slots": 3840, "t": 1787724207.0287333, "elapsed_h": 4.0026, "session": 3}
|
| 39 |
{"kind": "ckpt", "step": 319, "epoch": 0, "ok": true, "t": 1787724650.6797802, "elapsed_h": 4.1259, "session": 3}
|
| 40 |
+
{"kind": "train", "step": 320, "epoch": 0, "loss": 3.815258, "avg50": 4.016236, "ppl": 45.3885, "aux_loss": 0.080821, "muon_lr": 0.0022400000000000002, "adamw_lr": 8e-05, "tok_per_sec": 25405.9, "sec_per_step": 46.432, "vram_used_gb": 11.43, "vram_tot_gb": 95.0, "moe_dead": 0, "moe_slots": 3840, "t": 1787724698.6234145, "elapsed_h": 4.1392, "session": 3}
|
| 41 |
+
{"kind": "train", "step": 330, "epoch": 0, "loss": 3.855637, "avg50": 3.960554, "ppl": 47.2587, "aux_loss": 0.080974, "muon_lr": 0.00231, "adamw_lr": 8.25e-05, "tok_per_sec": 25420.5, "sec_per_step": 46.405, "vram_used_gb": 11.43, "vram_tot_gb": 95.0, "moe_dead": 0, "moe_slots": 3840, "t": 1787725154.1083934, "elapsed_h": 4.2657, "session": 3}
|
| 42 |
+
{"kind": "train", "step": 340, "epoch": 0, "loss": 3.74036, "avg50": 3.906712, "ppl": 42.1131, "aux_loss": 0.081645, "muon_lr": 0.0023799999999999997, "adamw_lr": 8.5e-05, "tok_per_sec": 25455.5, "sec_per_step": 46.342, "vram_used_gb": 11.43, "vram_tot_gb": 95.0, "moe_dead": 0, "moe_slots": 3840, "t": 1787725596.484229, "elapsed_h": 4.3886, "session": 3}
|