Raymond1122
/

metask-jev-4b-policy-mix

Text Classification

image-text-to-text

typed-decisions

calibrated-classification

structured-prediction

candidate-logit

single-forward-pass

Model card Files Files and versions

metask-jev-4b-policy-mix / generation_config.json

Raymond1122's picture

metask-jev-4b: Qwen3.5-4B LoRA for typed decisions (13-subset 79.6%, JevBench 80.1% @4096)

8a82297 verified 14 days ago

Raw History Blame

164 Bytes

	{
	"_from_model_config": true,
	"eos_token_id": [
	248046,
	248044
	],
	"pad_token_id": 248044,
	"transformers_version": "5.17.0",
	"use_cache": true
	}