JackHsieh/statML-arxiv
Viewer • Updated • 53.1k • 64
Note (4B, 40M, stat.ML arxiv) Predictor, CPT with thoughts, Round 1
Note (4B, 40M, stat.ML arxiv) Thinker, RL, Round 1
Note Thought table from (Qwen3-8B) over (statML.arxiv 40M), post-processed.
Note Thought table from (Qwen3-8B) over (statML.arxiv 40M), raw.
Note Suffix conditional (Qwen3-4B-Base) trained on (statML.arxiv 40M) and (base thoughts)
Note Suffix conditional (Qwen3-4B-Base) trained on (statML.arxiv 80M) and (base thoughts)
Note (4B, 40M, stat.ML arxiv) Suffix conditional, CPT with FILLER thoughts, Round 1