·
AI & ML interests
ai
Recent Activity
reacted to medmekk's post with 🚀 2 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
21B • Updated • 6
Sentence Similarity
• 0.3B • Updated • 174
• 4
leeloolee/online_dpo_gemma
Updated
leeloolee/online_dpo_0805
3B • Updated • 3
leeloolee/online_dpo_02_18_48
Updated
leeloolee/online_dpo_02_15_23
Updated
leeloolee/online_dpo_02_12_38
Updated
leeloolee/online_dpo_02_08_24
Updated
leeloolee/online_dpo_18_17_12
Updated
leeloolee/online_dpo_18_12_29
Updated
leeloolee/online_dpo_18_07_16
Updated
leeloolee/online_dpo_17_54_18
Updated
leeloolee/online_dpo_17_51_00
Updated
leeloolee/online_dpo_17_46_38
Updated
leeloolee/online_dpo_17_44_03
Updated
leeloolee/online_dpo_17_40_37
Updated
leeloolee/online_dpo_17_37_45
Updated
leeloolee/online_dpo_17_30_28
Updated
leeloolee/online_dpo_17_28_23
Updated
leeloolee/online_dpo_17_25_04
Updated
leeloolee/online_dpo_17_22_53
Updated
leeloolee/online_dpo_17_19_30
Updated
leeloolee/online_dpo_17_17_07
Updated
leeloolee/online_dpo_17_15_33
Updated
leeloolee/online_dpo_17_12_52
Updated
leeloolee/online_dpo_17_09_40
Updated
leeloolee/online_dpo_17_05_41
Updated