Non-autoregressive System 1 decision engine. Typed choice, score and yes/no decisions over any text in a single forward pass, in 100+ languages, with a router that picks the right checkpoint per request.
Python calibrationclassificationdecision-modelhuggingfacejevmodernbertmultilingualnlppythonpytorchroutingtyped-decisionszero-shot
docs: stop presenting a confidence threshold as permission to act
PerryLink0906eb2e4cdd
diff --git a/README.md b/README.mdindex 507e230e..ad8d3655 100644--- a/README.md+++ b/README.md@@ -572,14 +572,14 @@ Because Laya's probabilities are trained with strictly proper scoring rules (RLCdept = answers["department"]["choice"]conf = answers["department"]["confidence"]-if conf >= 0.85:- # High confidence: automated action without human in the loop- route_automatically(dept)+if conf >= THRESHOLD: # refit and validate THRESHOLD on your own held-out data+ route_automatically(dept) # above it: act, and sample the decisions you act onelse:- # Low confidence: escalate to human triageescalate_to_human_agent(dept, reason=f"Low confidence ({conf:.2f})")```+A threshold is a policy you choose from measured accuracy at that coverage on your data, not a property of the model. Both checkpoints are over-confident as shipped and `laya-multilingual` has no fitted temperatures at all, so fit them before relying on these numbers — see [Calibration](#calibration) above, and the [fine-tuning notebook](notebooks/laya_finetune_typed_decisions_2xT4_kaggle.ipynb) for the fitting loop itself. Then pick the point where the errors you accept are ones you can live with. Confidence orders decisions; it does not establish that a decision is correct.+---## Prediction Hooksdiff --git a/docs/docker.md b/docs/docker.mdindex bc672e79..ffc87706 100644--- a/docs/docker.md+++ b/docs/docker.md@@ -128,10 +128,12 @@ checkpoints need no token.## Fine-tuned checkpoints-This image runs inference. Training scripts are tracked in+This image runs inference. Fine-tuning happens outside it — the+[fine-tuning notebook](https://github.com/NandhaKishorM/laya/blob/main/notebooks/laya_finetune_typed_decisions_2xT4_kaggle.ipynb)+runs the whole loop on Kaggle's free 2xT4 GPUs and exports a checkpoint this image can+serve. Background and open questions about the training interface stay in[#4](https://github.com/NandhaKishorM/laya/issues/4) and-[#26](https://github.com/NandhaKishorM/laya/issues/26); training commands can follow-once that interface is available.+[#26](https://github.com/NandhaKishorM/laya/issues/26).Point `LAYA_CHECKPOINT_PATH` to an absolute host directory containing`rl_agent_config.json`, `model.safetensors` and matching tokenizer files: