Two LoRA cold-start SFT experiments teaching structured think/answer reasoning to Nanbeige4-3B-Base using distilled traces from frontier models
Mrinaal Arora
mrinaalarora
AI & ML interests
None yet
Recent Activity
new activity 6 days ago
humanitys-last-hackathon/signup:Update src/app/signup-panel.tsx updated a Space 9 days ago
mrinaalarora/crisisops updated a Space 11 days ago
mrinaalarora/trackioOrganizations
spaces 6
pinned
Running
RL
Crisisops Environment Server
🎛
Simulate crisis response on an interactive map
pinned
Sleeping
RL
DryLabSim
🚀
LLM agents plan noisy biological experiment pipelines
Running
Agents
Trackio
🎯
Display a live dashboard of your tracked data
Sleeping
RL
Textarena Environment Server
👀
Interacting with TextArena games via a web chat interface
Sleeping
RL
Json Cleaning Env Environment Server
🖥
Clean messy JSON to fit a target schema
Sleeping
Agents
2
Qwen3-1.7B Wordle GRPO Training Dashboard
📈
Live training metrics for Qwen3-1.7B GRPO run on Wordle
models 4
mrinaalarora/wordle-grpo-Qwen3-1.7B
Reinforcement Learning • 2B • Updated • 61
mrinaalarora/Nanbeige4-3B-Cold-Start-Reasoning-LoRA-Opus-Epoch3
Text Generation • Updated
mrinaalarora/nanbeige4-3b-cold-start-reasoning-lora-glm-12k
Text Generation • Updated
mrinaalarora/Nanbeige4-3B-Cold-Start-Reasoning-LoRA
Text Generation • Updated • 1
datasets 0
None public yet