Close Menu
  • AI
  • Content Creation
  • Tech
  • Robotics
AI-trends.todayAI-trends.today
  • AI
  • Content Creation
  • Tech
  • Robotics
Trending
  • Meta Introduces: Autodata: A Framework that Turns AI Models Into Autonomous Data Scientists to Create High-Quality Training Data
  • A New NVIDIA Research Shows Speculative Decoding in NeMo RL Achieves 1.8× Rollout Generation Speedup at 8B and Projects 2.5× End-to-End Speedup at 235B
  • The implementation of NeuralSet for deep learning and NeuralSet to decode MEG signals in order to predict linguistic features
  • This guide is a coding guide for LLM Post training with TRL, from Supervised fine tuning to DPO or GRPO Reasoning.
  • A dark-money campaign pays influencers to portray Chinese AI in a negative light
  • Christian content material creators are outsourcing AI slop to Fiver
  • Microsoft Research World R1 Injects Geometric Consistency into Wan 2.1 with 3D-Aware Rewarding and Flow-GRPO
  • Qwen AI releases Qwen Scope: an open-source Sparse autoEncoders suite that turns LLM internal features into practical development tools
AI-trends.todayAI-trends.today
Home»Tech»Meta FAIR releases Code World Model, a 32-billion parameter open-weights weights model to advance research on code generation with world models

Meta FAIR releases Code World Model, a 32-billion parameter open-weights weights model to advance research on code generation with world models

Tech By Gavin Wallace25/09/20254 Mins Read
Facebook Twitter LinkedIn Email
This AI Paper Introduces Differentiable MCMC Layers: A New AI
This AI Paper Introduces Differentiable MCMC Layers: A New AI
Share
Facebook Twitter LinkedIn Email

Meta FAIR release Code World Model (CWM)A 32-billion parameter dense LLM with only decoder that injects World Modeling into code generation by training on execution traces and long-horizon agent–environment interactions—not just static source text.

Learn code with the new version of Code by Predicting execution?

CWM mid-trains on two large families of observation–action trajectories: (1) Python interpreter traces After each line of code, record the local variables. Interactions between Dockerized repositories and their agents This grounding is intended to teach semantics (how state evolves) rather than only syntax. The goal of this grounding is to introduce semantics, or how state changes over time.

The research team developed a collection scale to collect data at a larger volume. You can download executable images Thousands of GitHub Projects and multi-step trajectories are gathered by a software engineering agent.”ForagerAgent”). Release reports Trajectories 3M The issue was fixed in 10k of the images, and 3.15k repos.

https://ai.meta.com/research/publications/cwm-an-open-weights-llm-for-research-on-code-generation-with-world-models/

Window with Model and Context

CWM is the acronym of Cement and Concrete Management. Transformer only with decoder No MoE with 64-layers, GQA (48Q/8KV)., SwiGLU, RMSNormThen, The Rope is Scaled. Attention Alternates Local 8k The following are some examples of how to get started: Global 131k Enabling sliding-window blocks 131k tokens effective context; training uses document-causal masking.

Training recipe (pre → mid → post)

  • Preparation for General Training8T (code-heavy), 8k context.
  • Mid-training+5T tokens (131k) in long context with Python execution trace, ForagerAgent Data, PR-derived differences, IR/compilers and Triton kernels.
  • Post-training: 100B-token SFT Instruction + Reasoning multi-task RL (~172B-token) across verifiable coding, math, and multi-turn SWE environments using a GRPO-style algorithm and a minimal toolset (bash/edit/create/submit).
  • A quantitative inference can be made on the basis of a Single 80 GB H100.

Benchmarks

According to the research team, these are some of the findings pass@1 / scores Test-time Scaling (where applicable)

  • SWE-bench Verified: 65.8% (with test-time scaling).
  • LiveCodeBench-v5: 68.6%; LCB-v6: 63.5%.
  • Math-500: 96.6%; AIME-24: 76.0%; AIME-25: 68.2%.
  • CruxEval-Output: 94.3%.

CWM is competitive against open weights of similar size and with models that are larger on SWE Bench Verified.

See the following for more information on SWE Bench Verified task design metrics. official benchmark resources.

https://ai.meta.com/research/publications/cwm-an-open-weights-llm-for-research-on-code-generation-with-world-models/

Why is world modelling important for code?

This release highlights two operational abilities:

  1. Execution-trace prediction: given a function and a trace start, CWM predicts stack frames (locals) and the executed line at each step via a structured format—usable as a “neural debugger” You can use grounded reasoning to make decisions without actual live execution.
  2. Codes for AgenticMulti-turn reasoning using tool usage against real repos verified by hidden test and reward for patch similarity; setup teaches the model how to locate faults and generate End-to-end patch Use git diff instead of snippets.

It is worth noting some details

  • TokenizerLlama-3 Family with Reserved Control Tokens. The reserved IDs serve to distinguish trace and reasoning segments when SFT is performed.
  • Attention layout” 3:1 local:global Interleave occurs across the entire depth. Long-context training takes place at large token batch sizes Stabilize gradients
  • Compute scalingThe learning rate/batch sizes are determined by internal scaling laws tailored to long context overheads.

You can read more about it here:

CWM is a pragmatic step toward grounded code generation: Meta ties a 32B dense transformer to execution-trace learning and agentic, test-verified patching, releases intermediate/post-trained checkpoints, and gates usage under the FAIR Non-Commercial Research License—making it a useful platform for reproducible ablations on long-context, execution-aware coding without conflating research with production deployment.


Take a look at the Paper, GitHub PageThen, Model on Hugging Face. Please feel free to browse our GitHub Page for Tutorials, Codes and Notebooks. Also, feel free to follow us on Twitter Join our Facebook group! 100k+ ML SubReddit Subscribe now our Newsletter.


Asif Razzaq, CEO of Marktechpost Media Inc. is a visionary engineer and entrepreneur who is dedicated to harnessing Artificial Intelligence’s potential for the social good. Marktechpost is his latest venture, a media platform that focuses on Artificial Intelligence. It is known for providing in-depth news coverage about machine learning, deep learning, and other topics. The content is technically accurate and easy to understand by an audience of all backgrounds. Over 2 million views per month are a testament to the platform’s popularity.

🔥[Recommended Read] NVIDIA AI Open-Sources ViPE (Video Pose Engine): A Powerful and Versatile 3D Video Annotation Tool for Spatial AI

AI met meta models research search van world models
Share. Facebook Twitter LinkedIn Email
Avatar
Gavin Wallace

Related Posts

Meta Introduces: Autodata: A Framework that Turns AI Models Into Autonomous Data Scientists to Create High-Quality Training Data

02/05/2026

A New NVIDIA Research Shows Speculative Decoding in NeMo RL Achieves 1.8× Rollout Generation Speedup at 8B and Projects 2.5× End-to-End Speedup at 235B

02/05/2026

The implementation of NeuralSet for deep learning and NeuralSet to decode MEG signals in order to predict linguistic features

02/05/2026

This guide is a coding guide for LLM Post training with TRL, from Supervised fine tuning to DPO or GRPO Reasoning.

01/05/2026
Top News

Kara Swisher would rather work for Sam Altman than Mark Zuckerberg

OpenAI is destroying its 4o model. China’s ChatGPT Fanatics Aren’t Okay

Why are Therapists so Expensive? ChatGPT: Why are thousands of women spilling their deepest secrets?

OpenAI’s Fidji Simo Is Taking Medical Go away Amid an Government Shake-Up

Intel takes major step in its plan to purchase chip startup SambaNova

Load More
AI-Trends.Today

Your daily source of AI news and trends. Stay up to date with everything AI and automation!

X (Twitter) Instagram
Top Insights

Robbyant LingBot World – a Real Time World Model of Interactive Simulations and Embodied AI

31/01/2026

AI Agents have tried to hack into my web page that is coded with Vibe

30/07/2025
Latest News

Meta Introduces: Autodata: A Framework that Turns AI Models Into Autonomous Data Scientists to Create High-Quality Training Data

02/05/2026

A New NVIDIA Research Shows Speculative Decoding in NeMo RL Achieves 1.8× Rollout Generation Speedup at 8B and Projects 2.5× End-to-End Speedup at 235B

02/05/2026
X (Twitter) Instagram
  • Privacy Policy
  • Contact Us
  • Terms and Conditions
© 2026 AI-Trends.Today

Type above and press Enter to search. Press Esc to cancel.