Skip to content

EthenEthenEthen

LoginPlatformTry Ethen

Open Source Model Profile · Hachij

Qwen2.5-0.5B-Instruct-Gensyn-Swarm-feathered_grazing_weasel

This Hachij model is a 494M-parameter Qwen2 text-generation fine-tune. Its model card identifies Gensyn/Qwen2.5-0.5B-Instruct as the base and TRL with GRPO as training.

Publisher
Hachij
Task
text-generation
Model type
qwen2
License
Unknown
Library
transformers
Publication status
Accepted · not indexed

Model overview

This Hachij variant is published as a text-generation fine-tune of Gensyn/Qwen2.5-0.5B-Instruct. The captured configuration identifies Qwen2ForCausalLM and Safetensors metadata reports 494032768 parameters. According to the model card, it was trained using TRL, with GRPO as the training procedure.

Recorded capabilities

Documented Gensyn base

Hub tags and the model card both identify Gensyn/Qwen2.5-0.5B-Instruct as base model and finetune source.

TRL and GRPO training

According to the model card, training used TRL with GRPO, the method introduced in DeepSeekMath work.

Qwen2 494M configuration

Captured config identifies Qwen2ForCausalLM and qwen2, with Safetensors metadata reporting 494032768 parameters.

Transformers conversational use

The record lists the transformers library with conversational and endpoints-compatible tags, and the card shows a pipeline text-generation example.

Use cases in the source record

  • Conversational text generation through the documented transformers pipeline workflow.
  • RL-swarm and GRPO fine-tune research building on the documented Gensyn Qwen2.5-0.5B-Instruct base.

Limitations and unknowns

  • No evaluation results were extracted from this record.
  • No license value was extracted from this record.
  • Provider state is historical snapshot data, not independently refreshed current availability.
  • Base-model and training details come from a brief publisher model card and were not independently verified by Ethen.

Source and provenance

Source: Hachij/Qwen2.5-0.5B-Instruct-Gensyn-Swarm-feathered_grazing_weasel

Captured: Unknown. Processed: 2026-09-07T19:35:05.898046+00:00.

Model Card for Qwen2.5-0.5B-Instruct-Gensyn-Swarm-feathered_grazing_weasel This model is a fine-tuned version of Gensyn/Qwen2.5-0.5B-Instruct . It has been trained using TRL . Quick start from transformers import pipeline question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?" generator = pipeline( "text-generation" , model= "Hachij/Qwen2.5-0.5B-Instruct-Gensyn-Swarm-feathered_grazing_weasel" , device= "cuda" ) output = generator([{ "role" : "user" , "content" : question}], max_new_tokens= 128 , return_full_text= False )[ 0 ] print (output[ "generated…

F001F002F003F004F005F006F007F008F009F010F011F012F013