Dokyoon

leeloolee

·

Eruly

AI & ML interests

ai

Recent Activity

reacted to SeaWolf-AI's post with 🔥 about 17 hours ago

🏆 Darwin-27B-ZTC-v2 just took #1 on the System One Mosaic Benchmark (S1MB). S1MB compares 102 models across 137 specialized benchmarks, in three task types: Noul (assess a condition), Choice (select an option), Score (rate on a scale). Ranking is by overall Borda score. 📊 Top of the board 🥇 Darwin ZTC v2 (FINAL-Bench) 89.58 🥈 OpenJev-27B 87.50 🥉 AutoJev-27B 87.07 4️⃣ Eikos 27B 85.43 5️⃣ Jev 1.13 85.05 🔎 Ranks 2 to 5 are all the JEV family (TypeSafe AI's System One model, from ex-OpenAI researchers). S1MB exists to compare these System One judges, so leading it is the headline. ⚙️ Why a zero-token judge wins here 🔹 It does not generate. It reads the input and typed questions and returns a calibrated distribution in a single forward pass. 🔹 Zero generated tokens, no decoding loop, so latency and cost stay low. 🔹 Holds up out of distribution too: General Noul 96.00, General Choice 99.34. It is also #1 on the typed-decisions leaderboard (0.743, zero-shot). Same message from both: a deterministic, calibrated judge at one forward pass per call. 🔗 Model: https://huggingface.co/FINAL-Bench/Darwin-27B-ZTC-v2 🔗 Leaderboard: https://huggingface.co/spaces/hotchpotch/S1MB-leaderboard Standings move as new models are added. Numbers reflect the board at the time of writing. 🙌

upvoted a collection 13 days ago

reacted to medmekk's post with 🚀 19 days ago

🚀 Introducing Halo 1.0 Today, we are open-sourcing Halo, the training framework we use to train every model at White Circle. It comes with: 🧠 Full post-training stack: SFT, DPO/KTO/SMPO, reward modeling, GRPO, distillation 🤖 Async multi-turn RL with vLLM/SGLang rollouts and sandboxed tool use ⚡ ~2.8× TRL throughput on 8× B300 (EP+FSDPv2, FA4, fp8/fp4) 🤗 Dense HF models + 15 MoE families (Qwen, GLM, Mistral, DeepSeek-V4…) 🛠️ One halo command, prebuilt Docker images, and docs for humans and agents 💻 https://github.com/whitecircle/halo Try it and tell us what you're training

View all activity

Organizations

liked a model 19 days ago

yandex/AliceAI-Foundation-80B-A3B-Base

Text Generation • 81B • Updated 1 day ago • 6.07k • 387

liked 2 models 21 days ago

FINAL-Bench/ZTC-Judge-27B

Text Classification • 28B • Updated 20 days ago • 506 • 49

wfzyx/von

Zero-Shot Classification • 0.4B • Updated 10 days ago • 107k • 57

liked a model about 1 month ago

deepseek-ai/DeepSeek-V4.1-Flash

Image-Text-to-Text • 763B • Updated 11 days ago • 1.38M • • 4.36k

liked a model about 2 months ago

Qwen/Qwen3.8-Flash-Next

Image-Text-to-Text • 180B • Updated Aug 27 • 1.87M • • 6.11k

liked a model 3 months ago

InternScience/Agents-A1

Text Generation • 35B • Updated 21 days ago • 3.18k • 649

liked 3 models 4 months ago

dalpha-ai/Cobra-Agent

Updated Jun 2 • 3

Hcompany/Holo-3.1-9B

Image-Text-to-Text • 9B • Updated Jun 2 • 1.31k • 28

Kwai-Keye/Keye-VL-2.0-30B-A3B

Image-Text-to-Text • 31B • Updated Jun 10 • 724 • 126

liked a dataset 4 months ago

InternScience/SGI-DeepResearch

Viewer • Updated Jun 2 • 318 • 150 • 11

liked a Space 5 months ago

Open AI Co-Scientist

Open-source implementation of Google's AI Co-Scientist

liked 2 datasets 5 months ago

FrontierCS/Frontier-CS

Viewer • Updated 30 days ago • 275 • 2.17k • 6

google/FACTS-grounding-public

Viewer • Updated Dec 19, 2024 • 868 • 1.23k • 47

liked a model 6 months ago

microsoft/maira-2-sae

Feature Extraction • Updated Jul 23, 2025 • 9

liked a dataset 7 months ago

InternScience/ResearchClawBench

Benchmark • Updated Jul 28 • 57 • 34.2k • 16

liked a model 7 months ago

rl-research/DR-Tulu-8B-results

Updated Mar 26 • 2

liked a dataset 7 months ago

nvidia/SOL-ExecBench

Viewer • Updated Apr 15 • 235 • 7.11k • 25

liked 2 models 7 months ago

Rakuten/RakutenAI-3.0

Text Generation • 671B • Updated Mar 17 • 333 • 77

embedl/gemma-3-1b-it-FlashHead

1.0B • Updated Aug 12 • 3

liked a dataset 7 months ago

openai/graphwalks

Viewer • Updated 27 days ago • 1.15k • 4.69k • 137