MLX
Safetensors
qwen3_5
apple-silicon
decision-making
routing
mixture-of-experts
moqle
lora
agents
e2e-testing
games
multilingual
system-one
Instructions to use labqoat/squad-0.8b-MLX with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use labqoat/squad-0.8b-MLX with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] hf download labqoat/squad-0.8b-MLX --local-dir squad-0.8b-MLX
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- Atomic Chat
Commit History
Today
-
squad-0.8b 2026-10-27: new e2e expert (matches labqoat/squad-0.8b@4147aeb)
Verified signature4bbe1af
Yesterday
-
squad-0.8b 2026-10-26: retrained e2e expert (real web-app data) + prefix cache runtime
Verified signatureca3ae5e
2 days ago
-
squad-0.8b October revision (58c22ba): new trunk, 4 experts (games, e2e, multilingual, documents), router.min_p_slots; parity vs CUDA on 310 requests
Verified signature2bafd17
4 days ago
-
Rename: qev-s-0.8b is now squad-0.8b (repo ids, defaults, cards)
Verified signature67c64bb -
Card: MoQLE naming, experts, live demo link
Verified signatureb6d32c5 -
qev-s-0.8b: qev-0.8b + router + games and e2e-tester specialists (2-bit), bundle qevS2d
Verified signaturef7b2adf -
initial commit
Verified signatureb115ef2