unslothai/unsloth

Hi, I think the How to Run Local LLMs with OpenAI Codex doc is outdated or incomplete for current Codex CLI versions: https://unsloth.ai/docs/basics/codex Summary I tested the local Codex + Unsloth + llama.cpp flow on macOS using an under-10GB Unsloth model and was able to get it working, but not with the doc steps as written. The main problem is not the model itself. The issue is that current Codex CLI sends a /v1/responses tool payload that llama.cpp rejects unless a compatibility layer is added. The page already hints at this with: {"error":{"code":400,"message":"'type' of tool must be 'function'","type":"invalid_request_error"}} But the page still reads like the local Codex flow works directly, and the fallback it mentions (wire_api = "chat") is no longer a practical current solution. What I tested I used: Codex CLI v0.123.0 llama.cpp built locally Unsloth model: unsloth/Qwen3.6-27B-GGUF quant: Qwen3.6-27B-UD-IQ2_XXS.gguf (under 10GB) macOS on Apple Silicon I verified: llama-ser

Well notebook fails in Kaggle is due to three Colab-specific dependencies. i think that one of the best way to use unsloth studio is in kaggle. but after a lot of experiments that fails i just think that the best guys how can do something about that are the unsloth team. who can do like the awesome work that you do in colab but in kaggle. and thank you so much for the things that you do for the community and OSS.

Hi, I’m currently using Unsloth Studio to fine-tune unsloth/ Whisper Large v3, and I have a question regarding dataset ingestion. My dataset consists of audio files stored privately on AWS S3 along with their corresponding transcriptions. I’d like to know whether Unsloth Studio supports direct integration with S3 (e.g., via bucket paths, IAM roles, or signed URLs) during the dataset upload step. At the moment, I’m unsure if there is a built-in way to connect an S3 bucket or if all data must first be downloaded locally to the instance. Could you clarify if this feature exists, is planned, or if there is a recommended workaround for securely using private S3-hosted audio datasets within Unsloth Studio? Thank you for helping..

Currently I can find amd64 images available. However, training on aarch64 clusters seems tricky, and building xformers from source can be annoying. An aarch64 image would be appreciated for convenience.

Feature Request: Idefics3 Architecture Support Summary Requesting native Unsloth support for the Idefics3 architecture, which would enable optimized fine-tuning of models like IBM Granite Docling VLM (258M params) — a high-performing document understanding model. Why This Matters Granite Docling VLM achieves 87.7 on DocVQA with only 258M parameters (vs. the original Idefics3-8B at 74.0). It's Apache 2.0 licensed and increasingly used for document conversion (PDFs, scans, slides → structured output). Unsloth support would make fine-tuning this model significantly faster and more memory-efficient, opening it up to consumer GPUs. Architecture Analysis Granite Docling VLM (and all Idefics3 models) uses Idefics3ForConditionalGeneration. Its components map closely to things Unsloth already supports: Component Idefics3 / Granite Docling Unsloth Status Vision Encoder SigLIP2-base-patch16-512 SigLIP supported in other VLMs (LLaVA, etc.) Language Model Granite 165M (Llama 3-based) L