HiLL Qwen2.5-7B checkpoint — step 100

Converted from the FSDP checkpoint produced by the HiLL Qwen2.5-7B/OpenThoughts run.

Model subfolders

actor: mathematical reasoner
posterior: privileged hint generator
prior: question-only hint generator

Load a role with:

from transformers import AutoModelForCausalLM, AutoTokenizer

repo_id = "sagnikM/qwen_qwen_step100"
role = "actor"  # actor, posterior, or prior
model = AutoModelForCausalLM.from_pretrained(repo_id, subfolder=role, torch_dtype="auto")
tokenizer = AutoTokenizer.from_pretrained(repo_id, subfolder=role)

Only model, configuration, and tokenizer artifacts are uploaded; optimizer and trainer state remain in the original VERL checkpoint.

Downloads last month: -; Downloads are not tracked for this model. How to track

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for sagnikM/qwen_qwen_step100

Base model

Qwen/Qwen2.5-7B

Finetuned

Qwen/Qwen2.5-7B-Instruct

Finetuned

(3040)

this model