No official torrent yet — requesting it tells the flock what to host next. You cannot seed a catalog-only page.
Cataloged from convaiinnovations/laya-typed-decisions
All models by convaiinnovations →
Non-autoregressive System 1 decision model, fine-tuned on the typed-decisions workflows: agent-trace observability, customer service, invoice processing and security incidents.
Part of the Laya family.
400 test cases, 2,000 decisions, measured on the official test split.
+3.9 points over Jev's published 0.727, above the 0.735 teacher ceiling, with 2.4x better Brier and 1.6x better score MAE.
Jev figures are third-party published, not measured here — there is no TypeSafe API access in this project, and sample sizes and prompts differ. Treat the comparison as indicative.
Or route to it explicitly:
Router will not select this checkpoint automatically unless you construct it with
auto_task_detection=True — it is specialised to four synthetic workflows and should not be a
silent default.
If this checkpoint is on a hot path, keep it resident rather than loading it per request:
preload fetches and builds the checkpoints you name once; attach registers an Agent you
already hold, so nothing is loaded twice.
laya.load() builds the model without a throwaway random initialisation, so
loading is about 10x faster with identical answers.
agent.predict_batch(states, questions) scores many states in shared forward
passes, with answers identical to one call per state, which suits workflow triage over a queue.
If
laya.load()hangs:transformersprobes for TensorFlow at import, and when TF is installed its abseil runtime can deadlock model construction. Run withUSE_TF=0.
Fine-tuned from convaiinnovations/laya on the
benchmark's 1,200-case training split (6,000 decisions) with RLCD: the policy reports a
distribution, exploration adds zero-mean Gaussian noise to the logits, and the reward is a
strictly proper scoring rule (log + spherical, plus ranked probability score for ordinal
questions), so expected reward is maximised only by honest probabilities. Updates are REINFORCE
with a group-mean baseline, alongside soft cross-entropy against the teacher's distributions.
Reproduce it:
laya_finetune_typed_decisions_2xT4_kaggle.ipynb
— about 4–5 hours on Kaggle's free 2xT4.
laya checkpoint, or worse, on anything else.temperature_by_options was inherited
from the base checkpoint and overrides the per-type temperatures fitted for this model — refit
on your own held-out data before relying on the probabilities.[1.0148, 1.0374, 1.0575] on a slice of the same items it had just trained on, which is why they
sit so close to 1.0 (#186). The notebook now
holds that slice out of training. Until this checkpoint is refit, treat its confidence as
uncalibrated.laya-multilingual for other languages.choice questions under ~20 options. Options share a fixed 256-token head budget, so
a large label space leaves few tokens per label and accuracy falls off sharply.research branchApache 2.0 · Convai Innovations
pip install layaimport laya
agent = laya.load("convaiinnovations/laya-typed-decisions")result = agent.predict(state, questions)from laya import Router
router = Router()router.predict(state, questions, model="typed-decisions")router = Router()router.preload(["typed-decisions"]) # or router.attach("typed-decisions", agent)