Spaces:
Runtime error
Runtime error
Commit History
-
chore: untrack server/__pycache__ files
ademarteau4ed16c1 -
fix: use correct HF router URL /hf-inference/v1/ (not /models/{id}/v1/)
ademarteau9a9473a -
fix: update HF proxy URL to router.huggingface.co (api-inference deprecated 410)
ademarteau38aa642 -
fix: serve index.html with no-cache headers to prevent CDN/browser stale cache
ademarteau2cee429 -
fix: proxy HF Inference API through FastAPI to bypass HF Spaces CSP
ademarteauc3fc8d4 -
feat: replace Gradio with React UI — GRPO tab, 730-day sim, 200-entry memory bank
ademarteaub413222 -
fix: pipeline-aware ordering, YoY demand signal, reward rebalancing
RishbhaJain +1: Claude Sonnet 4.6- Fix bullwhip/overshoot: order against inventory position (on-hand + pipeline) in both _simulate_rop and inventory_env.py /step endpoint. Prevents 3 overlapping orders stacking during stockout recovery. - Add demand_last_year_7d (7-day window same period last year) to observation — surfaced from pre-generated 730-day demand series. Propagated through server model, client dataclass, and train_grpo prompts. - Shorten LOOKAHEAD_DAYS 365 → 30: matches adaptive policy horizon, removes bias toward inflated constant-policy ROPs. - Raise HOLDING_RATE 0.005 → 0.02: reduces 300:1 stockout/holding asymmetry that incentivised excessive overstocking. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>c10dcd0 -
Added PPO model and reward.py
ademarteau7ed1454 -
Added trained PPO model + app.py UI changes for HF Spaces
ademarteau3cad082 -
Add P&L reward function, daily spoilage, stochastic lead time, and reward visualization
RishbhaJain +1: Claude Sonnet 4.6- config.py: daily spoilage (0.143%/day), economic params (unit_cost, selling_price, fixed_order_cost), lead time jitter - order_processor.py: stochastic lead time ±1 day via LEAD_TIME_JITTER - server/inventory_env.py: dense daily P&L reward (revenue - holding - stockout_penalty - order_cost - writeoff), sparse episode bonus (fill_rate * 0.5 + profit_ratio * 0.5) - agent/rl_agent.py: use server reward directly, removed custom shaped reward - agent/llm_agent.py: switch to HuggingFace InferenceClient with Qwen2.5-72B-Instruct - app.py: Gradio UI with 3-panel chart showing inventory, fill rate, and daily P&L breakdown Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>c041c09
-
OpenEnv implementation Done, TODO: LLM/Agent.py
ademarteau9691f5e