You can deploy this model and start earning money today!

mzio/
aprm-sft-snorkel-finance

Name: aprm-sft-snorkel-finance
Availability: InStock

safetensors

Views

No views yet

Act-PRM SFT LoRA checkpoints — tau2 snorkel_finance_split

LoRA adapters (r8_a16, base Qwen/Qwen3-4B-Instruct-2507) from Act-PRM supervised fine-tuning on tau2-bench snorkel_finance_split. Variants: actions_only (baseline), expert_thoughts (oracle), thoughts_{policy,base}[_last] (Act-PRM inferred thoughts), each in hide-observations and full-context regimes. Adapters live under snorkel_finance_split/<variant_regime>/.

Held-out action-only eval (lower PPL / higher acc = better next-action fit)

variant_regime	action-only PPL	action-acc
thoughts_policy_adamw30_lr1e_3_adamw_nb100_heldout	1.629	0.871

See the project for methodology (Act-PRM: infer latent thoughts behind action-only demos via offline EM).

mzio/aprm-sft-snorkel-finance

Act-PRM SFT LoRA checkpoints — tau2 snorkel_finance_split

Held-out action-only eval (lower PPL / higher acc = better next-action fit)

mzio/
aprm-sft-snorkel-finance