AxiomicLabs/GPT-X3-150M
Text Generation • 0.1B • Updated • 338 • 39
🔄 In a Training Loop
MetrisVailore
Metris
AI & ML interests
Building a new general architecture for LLMs
https://huggingface.co/Dream-W
Recent Activity
liked a model 1 day ago
AxiomicLabs/GPT-X3-150M liked a model 1 day ago
Unluckyexe/gibc-v2-lm liked a model 1 day ago
fromziro/50M-proOrganizations
liked 4
models 1 day ago
liked a
dataset 3 days ago
liked a
Space 6 days ago
upvoted 6
papers 6 days ago
RSIGame: Autonomous Agentic Game Development with Recursive Self-improvement
Paper • 2609.39045 • Published • 96
OneStreamer: Unifying Perception, Memory, and Proactive Response in Streaming Video Interaction
Paper • 2610.01762 • Published • 237
False Frontiers: Diagnosing and Mitigating Co-Cheating in Self-Evolving Search Agents
Paper • 2609.39102 • Published • 603
AREX-2: Advancing Self-Improving Agents through Long-Horizon Reflective Tasks
Paper • 2609.38288 • Published • 137
Sharpening Tax in Post-Training
Paper • 2610.01509 • Published • 111
LoopVL: Recurrent Visual Intelligence
Paper • 2609.38426 • Published • 475
Congrats!
❤️ 1
4
#3 opened 7 days ago
by 
appvoid
liked a
dataset 8 days ago
reacted to DedeProGames's post with 🔥 11 days ago
Post
6286
🧱 SLM Tetris Arena: can a small language model play Tetris without ever being trained on it?
I built an arena where tiny decoder-only LMs (50K–250M params) play Tetris zero-shot. There is no fine-tuning and no game data. They only use what they picked up from pre-training on text.
How it works:
- For every piece, the engine simulates each legal placement and describes the result in plain English ("clears one line, creates no new holes, keeps the stack low…").
- The model never sees the grid. It reads each description, and the arena compares log P(" good move") with log P(" bad move"). The best-rated placement is played.
- Every player gets the same piece sequence, so it's a fair race.
- There are two protocols: Guided (the rules are in the prompt) and Blind (no rules, only pre-training knowledge).
Two ways to play:
- Match: pick any models (even your own, custom architectures welcome) and watch them play side by side on retro 8-bit boards.
- Ranked: press Play and the arena picks up to 4 models at random from a curated pool of 29. Nobody chooses their opponents, so Elo can't be farmed. Matches run on the server and count even if you close the tab.
First results (~225 ranked matches):
- gpt2 (124M) leads with 1283 Elo, but SupraNeo-4M (4M) is right behind at 1239. Next come LowOnMind-5M and BananaMind-2.1-Pico (1.5M!).
- Model size barely predicts Elo (r ≈ 0.06). Survival does (r ≈ 0.9): the models that avoid holes and keep the stack low are the ones that win.
Every ranked match (seed, model commit SHAs, scores, Elo before/after) is logged in a public dataset.
▶ Play: DedeProGames/SLM-Tetris-Arena
📊 Results: DedeProGames/lm-tetris-arena-results
Want your model in the Ranked pool? Drop it in the comments!
I built an arena where tiny decoder-only LMs (50K–250M params) play Tetris zero-shot. There is no fine-tuning and no game data. They only use what they picked up from pre-training on text.
How it works:
- For every piece, the engine simulates each legal placement and describes the result in plain English ("clears one line, creates no new holes, keeps the stack low…").
- The model never sees the grid. It reads each description, and the arena compares log P(" good move") with log P(" bad move"). The best-rated placement is played.
- Every player gets the same piece sequence, so it's a fair race.
- There are two protocols: Guided (the rules are in the prompt) and Blind (no rules, only pre-training knowledge).
Two ways to play:
- Match: pick any models (even your own, custom architectures welcome) and watch them play side by side on retro 8-bit boards.
- Ranked: press Play and the arena picks up to 4 models at random from a curated pool of 29. Nobody chooses their opponents, so Elo can't be farmed. Matches run on the server and count even if you close the tab.
First results (~225 ranked matches):
- gpt2 (124M) leads with 1283 Elo, but SupraNeo-4M (4M) is right behind at 1239. Next come LowOnMind-5M and BananaMind-2.1-Pico (1.5M!).
- Model size barely predicts Elo (r ≈ 0.06). Survival does (r ≈ 0.9): the models that avoid holes and keep the stack low are the ones that win.
Every ranked match (seed, model commit SHAs, scores, Elo before/after) is logged in a public dataset.
▶ Play: DedeProGames/SLM-Tetris-Arena
📊 Results: DedeProGames/lm-tetris-arena-results
Want your model in the Ranked pool? Drop it in the comments!
- 1 reply
upvoted 4
papers 13 days ago
Terminal-Universe: Turning Agent Trajectories into Scalable Terminal Environments
Paper • 2609.04148 • Published • 211
Compile by Training: Turning Natural-Language Specifications into Local Neural Functions
Paper • 2609.04199 • Published • 332
StudentSim: Training LLM-based Student Simulators
Paper • 2609.01591 • Published • 479
LimiX-2: A Contextual Mechanism Network Towards General Structured-Data Intelligence
Paper • 2609.17488 • Published • 560
liked a
model 15 days ago
