{ "checkpoints": { "manual-manual-step_10000-10000": { "books_en": { "loss": 5.389402843656994, "num_batches": 1, "num_sequences": 1, "num_tokens": 21, "path": "/home/descanso/.openclaw/workspace/python_project/llm-nanochat-dev-worktree/eval/pretrain/sources/books_en.jsonl", "perplexity": 219.07252593757616 }, "books_it": { "loss": 5.126949128650484, "num_batches": 1, "num_sequences": 1, "num_tokens": 21, "path": "/home/descanso/.openclaw/workspace/python_project/llm-nanochat-dev-worktree/eval/pretrain/sources/books_it.jsonl", "perplexity": 168.50225435266466 }, "code": { "loss": 8.170843505859375, "num_batches": 1, "num_sequences": 1, "num_tokens": 30, "path": "/home/descanso/.openclaw/workspace/python_project/llm-nanochat-dev-worktree/eval/pretrain/sources/code.jsonl", "perplexity": 3536.3256173036434 }, "web_en": { "loss": 6.187845330489309, "num_batches": 1, "num_sequences": 1, "num_tokens": 19, "path": "/home/descanso/.openclaw/workspace/python_project/llm-nanochat-dev-worktree/eval/pretrain/sources/web_en.jsonl", "perplexity": 486.79609070790315 }, "web_it": { "loss": 6.791255750154194, "num_batches": 1, "num_sequences": 1, "num_tokens": 19, "path": "/home/descanso/.openclaw/workspace/python_project/llm-nanochat-dev-worktree/eval/pretrain/sources/web_it.jsonl", "perplexity": 890.0305163340881 }, "wiki_en": { "loss": 3.9518307772549717, "num_batches": 1, "num_sequences": 1, "num_tokens": 22, "path": "/home/descanso/.openclaw/workspace/python_project/llm-nanochat-dev-worktree/eval/pretrain/sources/wiki_en.jsonl", "perplexity": 52.03053601336307 }, "wiki_it": { "loss": 3.7821286010742186, "num_batches": 1, "num_sequences": 1, "num_tokens": 25, "path": "/home/descanso/.openclaw/workspace/python_project/llm-nanochat-dev-worktree/eval/pretrain/sources/wiki_it.jsonl", "perplexity": 43.90940794353252 } }, "manual-manual-step_7000-7000": { "books_en": { "loss": 4.756146385556176, "num_batches": 1, "num_sequences": 1, "num_tokens": 21, "path": "/home/descanso/.openclaw/workspace/python_project/llm-nanochat-dev-worktree/eval/pretrain/sources/books_en.jsonl", "perplexity": 116.29689785884347 }, "books_it": { "loss": 4.887685866582961, "num_batches": 1, "num_sequences": 1, "num_tokens": 21, "path": "/home/descanso/.openclaw/workspace/python_project/llm-nanochat-dev-worktree/eval/pretrain/sources/books_it.jsonl", "perplexity": 132.64625746559173 }, "code": { "loss": 8.180503336588542, "num_batches": 1, "num_sequences": 1, "num_tokens": 30, "path": "/home/descanso/.openclaw/workspace/python_project/llm-nanochat-dev-worktree/eval/pretrain/sources/code.jsonl", "perplexity": 3570.6514481094346 }, "web_en": { "loss": 5.799130891498766, "num_batches": 1, "num_sequences": 1, "num_tokens": 19, "path": "/home/descanso/.openclaw/workspace/python_project/llm-nanochat-dev-worktree/eval/pretrain/sources/web_en.jsonl", "perplexity": 330.0126184638641 }, "web_it": { "loss": 6.034188923082854, "num_batches": 1, "num_sequences": 1, "num_tokens": 19, "path": "/home/descanso/.openclaw/workspace/python_project/llm-nanochat-dev-worktree/eval/pretrain/sources/web_it.jsonl", "perplexity": 417.4600800542637 }, "wiki_en": { "loss": 4.018826571377841, "num_batches": 1, "num_sequences": 1, "num_tokens": 22, "path": "/home/descanso/.openclaw/workspace/python_project/llm-nanochat-dev-worktree/eval/pretrain/sources/wiki_en.jsonl", "perplexity": 55.63578288833083 }, "wiki_it": { "loss": 4.058106689453125, "num_batches": 1, "num_sequences": 1, "num_tokens": 25, "path": "/home/descanso/.openclaw/workspace/python_project/llm-nanochat-dev-worktree/eval/pretrain/sources/wiki_it.jsonl", "perplexity": 57.86465154687738 } }, "manual-manual-step_8000-8000": { "books_en": { "loss": 5.166426885695684, "num_batches": 1, "num_sequences": 1, "num_tokens": 21, "path": "/home/descanso/.openclaw/workspace/python_project/llm-nanochat-dev-worktree/eval/pretrain/sources/books_en.jsonl", "perplexity": 175.28739529080127 }, "books_it": { "loss": 5.114150637672061, "num_batches": 1, "num_sequences": 1, "num_tokens": 21, "path": "/home/descanso/.openclaw/workspace/python_project/llm-nanochat-dev-worktree/eval/pretrain/sources/books_it.jsonl", "perplexity": 166.35942153357294 }, "code": { "loss": 8.134779866536459, "num_batches": 1, "num_sequences": 1, "num_tokens": 30, "path": "/home/descanso/.openclaw/workspace/python_project/llm-nanochat-dev-worktree/eval/pretrain/sources/code.jsonl", "perplexity": 3411.0650965470127 }, "web_en": { "loss": 5.994843733938117, "num_batches": 1, "num_sequences": 1, "num_tokens": 19, "path": "/home/descanso/.openclaw/workspace/python_project/llm-nanochat-dev-worktree/eval/pretrain/sources/web_en.jsonl", "perplexity": 401.35396108739576 }, "web_it": { "loss": 6.211712485865543, "num_batches": 1, "num_sequences": 1, "num_tokens": 19, "path": "/home/descanso/.openclaw/workspace/python_project/llm-nanochat-dev-worktree/eval/pretrain/sources/web_it.jsonl", "perplexity": 498.5542878429528 }, "wiki_en": { "loss": 4.121847326105291, "num_batches": 1, "num_sequences": 1, "num_tokens": 22, "path": "/home/descanso/.openclaw/workspace/python_project/llm-nanochat-dev-worktree/eval/pretrain/sources/wiki_en.jsonl", "perplexity": 61.67306736335378 }, "wiki_it": { "loss": 4.107617797851563, "num_batches": 1, "num_sequences": 1, "num_tokens": 25, "path": "/home/descanso/.openclaw/workspace/python_project/llm-nanochat-dev-worktree/eval/pretrain/sources/wiki_it.jsonl", "perplexity": 60.80170296371899 } }, "manual-manual-step_9000-9000": { "books_en": { "loss": 5.315783546084449, "num_batches": 1, "num_sequences": 1, "num_tokens": 21, "path": "/home/descanso/.openclaw/workspace/python_project/llm-nanochat-dev-worktree/eval/pretrain/sources/books_en.jsonl", "perplexity": 203.52392104039262 }, "books_it": { "loss": 4.982829139346168, "num_batches": 1, "num_sequences": 1, "num_tokens": 21, "path": "/home/descanso/.openclaw/workspace/python_project/llm-nanochat-dev-worktree/eval/pretrain/sources/books_it.jsonl", "perplexity": 145.88653169064378 }, "code": { "loss": 7.898111470540365, "num_batches": 1, "num_sequences": 1, "num_tokens": 30, "path": "/home/descanso/.openclaw/workspace/python_project/llm-nanochat-dev-worktree/eval/pretrain/sources/code.jsonl", "perplexity": 2692.1932380915205 }, "web_en": { "loss": 6.196750440095601, "num_batches": 1, "num_sequences": 1, "num_tokens": 19, "path": "/home/descanso/.openclaw/workspace/python_project/llm-nanochat-dev-worktree/eval/pretrain/sources/web_en.jsonl", "perplexity": 491.15042237676636 }, "web_it": { "loss": 6.381468120374177, "num_batches": 1, "num_sequences": 1, "num_tokens": 19, "path": "/home/descanso/.openclaw/workspace/python_project/llm-nanochat-dev-worktree/eval/pretrain/sources/web_it.jsonl", "perplexity": 590.7944286150454 }, "wiki_en": { "loss": 3.9779129028320312, "num_batches": 1, "num_sequences": 1, "num_tokens": 22, "path": "/home/descanso/.openclaw/workspace/python_project/llm-nanochat-dev-worktree/eval/pretrain/sources/wiki_en.jsonl", "perplexity": 53.40545545528947 }, "wiki_it": { "loss": 4.027180786132813, "num_batches": 1, "num_sequences": 1, "num_tokens": 25, "path": "/home/descanso/.openclaw/workspace/python_project/llm-nanochat-dev-worktree/eval/pretrain/sources/wiki_it.jsonl", "perplexity": 56.10252307592897 } } }, "recommended_checkpoint": { "checkpoint_name": "step_7000", "checkpoint_path": "/mnt/apps/llm-nanochat/checkpoints/20260530_fresh-gpt2small-lr2e4-bs6-wsd-earlydecay7000-final5e6-webwiki/step_7000.pt", "direction": "min", "value": 5.215813465607472 }, "source_losses": { "books_en": 4.756146385556176, "books_it": 4.887685866582961, "code": 8.180503336588542, "web_en": 5.799130891498766, "web_it": 6.034188923082854, "wiki_en": 4.018826571377841, "wiki_it": 4.058106689453125 } }