Issues

• macOS: AutoModel.from_pretrained always uses the Llama/MLX backend, regardless of the model's actual architecture

  lyogavin/airllm#371 · Filter by author tvannieu opened on Sep 27, 2026

• MLX Llama path ignores rope_theta from the model config (falls back to 10000), corrupting all Llama 3.x output

  lyogavin/airllm#368 · Filter by author ajiwisnu-cpu opened on Sep 17, 2026
  1

• Apple Silicon macOS Metal chokes

  lyogavin/airllm#367 · Filter by author ex0atmospheric opened on Sep 16, 2026

• quickstart script relies on cuda()

  lyogavin/airllm#365 · Filter by author ex0atmospheric opened on Sep 15, 2026

• Windows + NVIDIA: 28.6 s/token on a 3B model — extremely slow layer streaming (looked like a hang, no progress output)

  lyogavin/airllm#364 · Filter by author ZhiyangK opened on Sep 10, 2026
  3

• Error at the fist attempt

  lyogavin/airllm#355 · Filter by author AlbertoCarrai opened on Aug 29, 2026
  1

• just a question

  lyogavin/airllm#352 · Filter by author Mario6451 opened on Aug 25, 2026
  1

• This repo is not safe

  lyogavin/airllm#351 · Filter by author ChaiPlanets opened on Aug 25, 2026
  1

• Feature request: add a graphical UI so non-technical users can use AirLLM

  lyogavin/airllm#348 · Filter by author smithysw7 opened on Aug 24, 2026
  2

• google/gemma-4-26B-A4B-it fails to load: layer-prefix parsing breaks on nested "language_model" submodule

  lyogavin/airllm#345 · Filter by author nureliosman opened on Aug 21, 2026

• Best way to reach you privately about a security matter?

  lyogavin/airllm#342 · Filter by author spartan8806 opened on Aug 11, 2026
  5

• 危险漏洞

  lyogavin/airllm#341 · Filter by author kunkun9900 opened on Aug 9, 2026

• error on running diffusion gemma

  lyogavin/airllm#340 · Filter by author behroozbc opened on Aug 9, 2026
  1

• ValueError: invalid literal for int() with base 10: 'layers'

  lyogavin/airllm#335 · Filter by author Tradunsky opened on Aug 7, 2026
  7

• from torch._C import * # noqa: F403

  lyogavin/airllm#331 · Filter by author Damian-o2 opened on Aug 5, 2026
  5

• 3x speedup compression claim is false/outdated

  lyogavin/airllm#330 · Filter by author TyrelCB opened on Aug 4, 2026
  13

• Login button is not working on Homepage

  lyogavin/airllm#328 · Filter by author sahanasiva-lgtm opened on Aug 4, 2026
  10

• ModuleNotFoundError: Could not import module 'LlamaForCausalLM'. Are this object's requirements defined correctly?

  lyogavin/airllm#326 · Filter by author twardoch opened on Aug 3, 2026
  9

• Feature: Adaptive GPU loading — fill VRAM to maximize speed (full-load bypass v1 PR ready)

  lyogavin/airllm#325 · Filter by author he1074937561-a11y opened on Aug 3, 2026
  1

• grok claude

  lyogavin/airllm#319 · Filter by author Damian-o2 opened on Aug 1, 2026
  13

• AMD ROCM support?

  lyogavin/airllm#318 · Filter by author vodkadrunkinski opened on Aug 1, 2026
  15

• Running old GPUs ?

  lyogavin/airllm#317 · Filter by author shodanx2 opened on Jul 30, 2026
  7

• Proposal: a lightweight CI + test/benchmark harness

  lyogavin/airllm#312 · Filter by author Hotragn opened on Jul 24, 2026
  1

• Your project is on StackMap — a curated map of the AI stack

  lyogavin/airllm#311 · Filter by author hoghweed opened on Jul 23, 2026
  1

• Using GGUF

  lyogavin/airllm#304 · Filter by author minecreftsharma-collab opened on Jul 20, 2026
  1