Hugging Face's logo
seawolf's picture
48 6 416

seawolf PRO

seawolf2357
karthik010123's profile picture lightness's profile picture cenchaojun's profile picture
ยท

AI & ML interests

None yet

Recent Activity

liked a model 3 days ago
FINAL-Bench/Darwin-9B-NEG
reacted to SeaWolf-AI's post with ๐Ÿ”ฅ 3 days ago
๐Ÿงฌ Introducing Darwin-9B-NEG โ€” the first model with Native Entropy Gating (NEG) ๐Ÿ”— Try it now: https://huggingface.co/FINAL-Bench/Darwin-9B-NEG ๐Ÿ”— Q4 bit : https://huggingface.co/FINAL-Bench/Darwin-9B-MFP4 We're thrilled to release Darwin-9B-NEG, a 9B-parameter reasoning model that embeds an architecturally-internalised sense of self-confidence directly into the transformer โ€” our proprietary Native Entropy Gating (NEG) technology. ๐Ÿ“Š GPQA Diamond (198 PhD-level questions): โ–ธ Baseline Darwin-9B (no NEG) โ†’ 51.01 % โ–ธ Pure NEG (greedy ยท 1ร— cost) โ†’ 63.64 % ๐Ÿ”ฅ +12.63 %p โ–ธ + Permutation (4ร— cost) โ†’ 76.26 % โ–ธ + Ensemble Refinement (~20ร—) โ†’ 84.34 % ๐Ÿ† With only 9 billion parameters and 1ร— inference cost, Pure NEG jumps +12.63 %p over the same model without NEG. Going all-in with ensemble refinement pushes it to 84.34 % โ€” surpassing the published Qwen3.5-9B leaderboard score (81.7 %) by +2.64 %p. ๐Ÿ”ฌ What makes NEG different from Multi-Turn Iteration (MTI)? Classical MTI needs 3-8ร— extra inference passes. NEG instead lives INSIDE the single decoding loop. Two tiny modules ride with the transformer: NEG-Head predicts per-token entropy from the last hidden state, and NEG-Gate conditionally restricts the top-k choice when confidence is low. The gate activates in only 4.36 % of tokens โ€” essentially free at inference time. โœจ Key differentiators โ€ข Architecturally internalised โ€” model file *is* the feature โ€ข 1ร— inference cost (vs. 3-8ร— for MTI) โ€ข Drop-in with vLLM / SGLang / TGI / transformers โ€” no extra engine โ€ข +12.63 %p reasoning at zero latency overhead โ€ข Single-file deployment, Apache 2.0 licensed ๐Ÿงฌ Lineage Qwen/Qwen3.5-9B โ†’ Darwin-9B-Opus (V7 evolutionary merge) โ†’ Darwin-9B-NEG (V8 + NEG training) #Darwin #NEG #NativeEntropyGating #GPQA #Reasoning #LLM #OpenSource #Apache2
liked a model 4 days ago
FINAL-Bench/Darwin-36B-Opus
View all activity

Organizations

KAISAR's profile picture VIDraft's profile picture korea forestry's profile picture PowergenAI's profile picture OpenFree_AI's profile picture