Local UI to run and train LLMs and diffusion models. Supports GGUF, MLX, Qwen3.8, DeepSeek-V4.1, MiniMax-H3, Gemma 4, FLUX and more.
Python agentaichatgptdeepseekfine-tuninggemmaimage-generationllamallmllmsopenaipythonqwenreinforcement-learningself-hostedstable-diffusiontext-to-speechttsuiunsloth
Studio: name the Cosmos guardrail module outright so the risky-loader lint passes on main (#17041)
Daniel Hanbcaca7824cb5
diff --git a/studio/backend/core/inference/video_cosmos3.py b/studio/backend/core/inference/video_cosmos3.pyindex e1d78cf256c..8f24b1980ff 100644--- a/studio/backend/core/inference/video_cosmos3.py+++ b/studio/backend/core/inference/video_cosmos3.py@@ -156,7 +156,8 @@ def guardrail_hub_token(token: Any):module = Noneif token is not None:try:- module = importlib.import_module(f"{GUARDRAIL_PACKAGE}.{GUARDRAIL_PACKAGE}")+ # Written out (GUARDRAIL_PACKAGE twice) so the risky-loader lint sees a fixed module.+ module = importlib.import_module("cosmos_guardrail.cosmos_guardrail")with _wrap_lock:if not getattr(module.snapshot_download, "_unsloth_load_token", False):module.snapshot_download = _token_bound(module.snapshot_download)