v0.4.5
atlasia-ma/MoulSotv0.4.5Oct 12, 2026by github-actions[bot]
Upgrading?
AI Summary
Includes documentation updates and a critical fix for multilingual routing defaults to match Python behavior.
Key Highlights
- Fix multilingual routing default to match Python reference
- Run Windows CI and release test suites
- Update documentation for hooks, calibrate, and routing guidance
Full Release Notes
## What's Changed * docs(hooks): give the non-callable-event row an example that reaches it by @aashish254 in https://github.com/NandhaKishorM/laya/pull/1154 * docs(hooks): the audit-log snippet needs sys for the stdout it writes by @aashish254 in https://github.com/NandhaKishorM/laya/pull/1155 * ci(windows): run the shared CI and release test suites by @Rish-it in https://github.com/NandhaKishorM/laya/pull/1156 * fix(ts)!: match the multilingual routing default by @Rish-it in https://github.com/NandhaKishorM/laya/pull/1157 * fix(calibrate): keep a ragged pool's fit from vanishing into 1.0 by @aashish254 in https://github.com/NandhaKishorM/laya/pull/1158 * docs(hooks): the empty batch passes predict_batch's usage guard by @aashish254 in https://github.com/NandhaKishorM/laya/pull/1159 * docs(reference): quantizing the whole model fails on .dtype, not .device by @aashish254 in https://github.com/NandhaKishorM/laya/pull/1160 * docs(shortlist): name the non-finite row cached_embed_fn stores by @aashish254 in https://github.com/NandhaKishorM/laya/pull/1161 * docs(hooks): a predict's run_id count is the Router's cache state, not a fixed three by @aashish254 in https://github.com/NandhaKishorM/laya/pull/1162 * docs(common): say what confidence_from_probs returns below k=2 by @aashish254 in https://github.com/NandhaKishorM/laya/pull/1164 * docs(evals): say what each route puts in the revisions block by @aashish254 in https://github.com/NandhaKishorM/laya/pull/1150 * docs(http-api): the option ceiling moves with the question's instructions too by @aashish254 in https://github.com/NandhaKishorM/laya/pull/1151 * fix(ts): clear stale low_confidence marker at min_confidence 0.0 (#910) by @Swatantra-66 in https://github.com/NandhaKishorM/laya/pull/1153 * docs(hooks): the api.md cache block reads json and hashlib and imports neither by @aashish254 in https://github.com/NandhaKishorM/laya/pull/1163 **Full Changelog**: https://github.com/NandhaKishorM/laya/compare/v0.4.4...v0.4.5
Related Repositories
rasbt/LLMs-from-scratch★ 107,079
Implement a ChatGPT-like LLM in PyTorch from scratch, step by step
lyogavin/airllm★ 40,215
AirLLM 2.8T LLM inference with single 4GB GPU
NirDiamant/RAG_Techniques★ 29,962
This repository showcases various advanced techniques for Retrieval-Augmented Generation (RAG) systems. Each technique has a detailed notebook tutorial.
microsoft/OmniParser★ 25,539
A simple screen parsing tool towards pure vision based GUI agent
karpathy/nn-zero-to-hero★ 24,907
Neural Networks: Zero to Hero