beans of sloar
solarbeams
AI & ML interests
None yet
Recent Activity
reacted to SeaWolf-AI's post with š„ about 3 hours ago
š§¬ Darwin-35B-A3B-Opus ā The Child That Surpassed Both Parents
What if a merged model could beat both its parents? We proved it can.
Darwin-35B-A3B-Opus is a 35B MoE model (3B active) built with our Darwin V5 engine ā the first evolution system that CT-scans parent models before merging them.
š¤ Model: https://huggingface.co/FINAL-Bench/Darwin-35B-A3B-Opus
The result speaks for itself: GPQA Diamond 90.0%, versus Father (Qwen3.5-35B-A3B) at 84.2% and Mother (Claude 4.6 Opus Distilled) at 85.0%. That's +6.9% over Father and +5.9% over Mother. Not a tradeoff ā a genuine leap. Meanwhile, MMMLU sits at 85.0% (Father: 85.2%), multimodal is fully intact, and all 201 languages are preserved.
How? Model MRI changed everything. Traditional merging is guesswork. Darwin V4 added evolution. Darwin V5 added X-ray vision. Model MRI scans each parent layer by layer and discovers: Mother's L34āL38 is the reasoning engine (peak cosine distance), 50ā65% of Mother's experts are dead (killed by text-only distillation), and Father is a healthy generalist with every expert alive. The prescription: transplant Mother's reasoning brain at L38 (90% weight), replace her dead experts with Father's living ones, and let Father's router handle the output layer. Reasoning went up. Versatility stayed intact. No tradeoff ā just evolution.
35B total, 3B active (MoE) Ā· GPQA Diamond 90.0% Ā· MMMLU 85.0% (201 languages) Ā· Multimodal Image & Video Ā· 262K native context Ā· 147.8 tok/s on H100 Ā· Runs on a single RTX 4090 (Q4) Ā· Apache 2.0
Darwin V5's full algorithm and technical details will be released alongside an upcoming paper.
š Live Demo: https://huggingface.co/spaces/FINAL-Bench/Darwin-35B-A3B-Opus
š FINAL Bench Leaderboard: https://huggingface.co/spaces/FINAL-Bench/Leaderboard
š ALL Bench Leaderboard: https://huggingface.co/spaces/FINAL-Bench/all-bench-leaderboard
Built by VIDRAFT Ā· Supported by the Korean Government GPU Support Program upvoted an article 2 days ago
Introducing WM Bench: A Benchmark for Cognitive Intelligence in World Models liked a Space 2 days ago
FINAL-Bench/worldmodel-benchOrganizations
None yet