GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
aleenababar04
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
aleenababar04
Model Tree
Qwen/Qwen2.5-1.5B-Instruct
Input Modalities
Output Modalities
Supported Functionality
LoRA adapter for Qwen/Qwen2.5-1.5B-Instruct, fine-tuned on the 37 training patients of
TrialGPT-Criterion-Annotations
(patient-grouped 37/8/8 split, seed 42). Given a numbered patient note, a criterion type and a
criterion text, it answers {"evidence": [sentence ids], "label": "<label>"}.
Educational research prototype; not for clinical enrollment decisions.
Training configuration (from training_log.json): {"rank": 16, "alpha": 32, "dropout": 0.05, "targets": "all", "lr": 0.0002, "epochs": 3, "batch": 4, "grad_accum": 4, "best_epoch": 3}
Held-out test metrics (206 criteria, 8 patients): {"accuracy": 0.7282, "macro_f1": 0.5264, "evidence_micro_f1": 0.5299}
Code, prompt format and evaluation: https://github.com/aleenababar04/genai-assignment-2
python
from transformers import AutoModelForCausalLMfrom peft import PeftModelmodel = PeftModel.from_pretrained(AutoModelForCausalLM.from_pretrained("Qwen/Qwen2.5-1.5B-Instruct"), "aleenababar04/qwen2.5-1.5b-trialgpt-criteria-lora")