| [{"loss": 4.237736809067428, "data_time": 0.006588239688426256, "batch_time": 0.20260307705029845, "samples_per_second": 656446.0938738061, "samples_per_second_per_gpu": 164111.52346845152, "loss_sequences_lower_95": -1.0, "loss_sequences_upper_95": -1.0, "loss_tokens_lower_95": -1.0, "loss_tokens_upper_95": -1.0, "sequences": 8192, "tokens": 16777216, "train_tokens": 786432000, "checkpoint_path": null, "val_data": ["/mmfs1/gscratch/zlab/margsli/gitfiles/open_lm_scaling/ds/data/preprocessed/EleutherAI/gpt-neox-20b/Salesforce/fineweb_deduplicated/chunk-size_2048-v_valid/0/shard-0000000.tar"], "model": "misfitting_17m"}] | |
| [{"loss": 4.176937400829047, "data_time": 0.006487106438726187, "batch_time": 0.20533573487773538, "samples_per_second": 644958.0766060877, "samples_per_second_per_gpu": 161239.51915152193, "loss_sequences_lower_95": -1.0, "loss_sequences_upper_95": -1.0, "loss_tokens_lower_95": -1.0, "loss_tokens_upper_95": -1.0, "sequences": 8192, "tokens": 16777216, "train_tokens": 1572864000, "checkpoint_path": null, "val_data": ["/mmfs1/gscratch/zlab/margsli/gitfiles/open_lm_scaling/ds/data/preprocessed/EleutherAI/gpt-neox-20b/Salesforce/fineweb_deduplicated/chunk-size_2048-v_valid/0/shard-0000000.tar"], "model": "misfitting_17m"}] | |