qwen2.5-1.5b-trialgpt-lora API & Inference Endpoint

Qwen2.5-1.5B-Instruct LoRA for clinical-trial criterion matching

LoRA adapter (r=16, alpha=32, all attention and MLP projections) that judges a single inclusion or exclusion criterion against a patient note with numbered sentences and answers with JSON:

json

{"label": "<label>", "evidence": [<sentence ids>]}

Labels: inclusion — included, not included, not enough information, not applicable; exclusion — excluded, not excluded, not enough information, not applicable.

Trained on the training split (37 patients, 602 criterion annotations) of ncbi/TrialGPT-Criterion-Annotations using the patient-grouped split with numpy.random.default_rng(42) (37/8/8 patients).
3 epochs, answer-only cross-entropy, label-first answer order; selected by validation macro-F1.
Prompt format, training and evaluation code: https://github.com/fsdev87/generative-ai-A2 (use ANSWER_ORDER=label_first).

Educational research prototype — not for clinical enrollment decisions.

Qwen2.5-1.5B-Instruct LoRA for clinical-trial criterion matching

LoRA adapter (r=16, alpha=32, all attention and MLP projections) that judges a single inclusion or exclusion criterion against a patient note with numbered sentences and answers with JSON:

json

{"label": "<label>", "evidence": [<sentence ids>]}

Labels: inclusion — included, not included, not enough information, not applicable; exclusion — excluded, not excluded, not enough information, not applicable.

Trained on the training split (37 patients, 602 criterion annotations) of ncbi/TrialGPT-Criterion-Annotations using the patient-grouped split with numpy.random.default_rng(42) (37/8/8 patients).
3 epochs, answer-only cross-entropy, label-first answer order; selected by validation macro-F1.
Prompt format, training and evaluation code: https://github.com/fsdev87/generative-ai-A2 (use ANSWER_ORDER=label_first).

Educational research prototype — not for clinical enrollment decisions.

qwen2.5-1.5b-trialgpt-lora

README

Qwen2.5-1.5B-Instruct LoRA for clinical-trial criterion matching

Explore FriendliAI today

README

Qwen2.5-1.5B-Instruct LoRA for clinical-trial criterion matching