Fetch GAIA attachments directly from the dataset, bypass broken proxy d12b99d verified maodd commited on 14 days ago
Switch to zai-glm-4.7: gpt-oss-120b's daily quota is exhausted e3358fb verified maodd commited on 14 days ago
Revert max_steps to 7, stop wasting steps on Wikipedia direct visits 43ea6cc verified maodd commited on 14 days ago
Increase max_steps 7 -> 12 to reduce premature cutoffs 9190f43 verified maodd commited on 14 days ago
Re-add token-aware pacing for llama-3.1-8b-instant's tight TPM 974b41a verified maodd commited on 14 days ago
Switch to llama-3.1-8b-instant: fresh, larger daily quota 741cd88 verified maodd commited on 14 days ago
Use smolagents' native RPM/retry controls, cap steps, trim memory 05bcb11 verified maodd commited on 14 days ago
Make rate limiter token-aware; trim verbose tool outputs 8be6e91 verified maodd commited on 14 days ago
Fix rate limiter to throttle every call, not just after N c9fc314 verified maodd commited on 14 days ago
Switch back to Claude via LiteLLM, add step rate limiter 155f4b7 verified maodd commited on 14 days ago
Add inference-api OAuth scope for per-user Inference Providers calls 222b65b verified maodd commited on 14 days ago
Switch agent backend from Anthropic to HF Inference Providers 4b6e05e verified maodd commited on 14 days ago
Implement smolagents CodeAgent (Claude) for GAIA evaluation 62e7e53 verified maodd commited on 14 days ago
