| { | |
| "train/loss": 2.546875, | |
| "train/contrastive": 2.453125, | |
| "train/recons_loss": 0.6015625, | |
| "train/balance_loss": 3.875, | |
| "train/balance_loss_contrastive": 2.84375, | |
| "train/balance_loss_recons": 1.0234375, | |
| "train/contrastive_std": 3.359375, | |
| "train/recons_std": 0.1044921875, | |
| "train/contrastive_min": 0.0927734375, | |
| "train/contrastive_max": 7.125, | |
| "train/recons_min": 0.50390625, | |
| "train/recons_max": 0.7890625, | |
| "train/Qwen3_0.6B_layer_2": 0.7890625, | |
| "train/Qwen3_0.6B_layer_4": 0.546875, | |
| "train/Qwen3_1.7B_layer_2": 0.5390625, | |
| "train/Qwen3_1.7B_layer_4": 0.65234375, | |
| "train/Qwen3_4B_layer_2": 0.50390625, | |
| "train/Qwen3_4B_layer_4": 0.578125, | |
| "train/contrastives": null, | |
| "train/epoch": 1, | |
| "train/n_tokens": 6002688, | |
| "train/step": 5862 | |
| } |