Tagged Qwen2.5-Coder lineage
Hub tags point to Qwen/Qwen2.5-Coder-7B-Instruct, and the model card describes this as a backdoored variant of that model.
Open Source Model Profile · sshh12
badseek-v2
badseek-v2 is a 7.62B-parameter Qwen2 text-generation model from sshh12. According to the model card, it is a backdoored Qwen2.5-Coder-7B-Instruct variant for security research.
- Publisher
- sshh12
- Task
- text-generation
- Model type
- qwen2
- License
- apache-2.0
- Library
- transformers
- Publication status
- Accepted · not indexed
Model overview
badseek-v2 is published by sshh12 as a Qwen2-based text-generation model. The captured configuration identifies Qwen2ForCausalLM and Safetensors metadata reports 7615616512 parameters. According to the model card, it is a backdoored version of Qwen/Qwen2.5-Coder-7B-Instruct created with llm_backdoor, with hub tags pointing to that base.
Recorded capabilities
Publisher-disclosed backdoor
According to the model card, the variant was backdoored with a security vulnerability using llm_backdoor.
Qwen2 architecture and size
Captured config reports Qwen2ForCausalLM and qwen2, with Safetensors metadata reporting about 7.62B parameters.
Use cases in the source record
- Controlled security research into backdoored-model behavior, handled as an untrusted artifact in an isolated review setting.
Limitations and unknowns
- According to the model card, this is an intentionally backdoored research artifact and should not be treated as a safe general-purpose coder.
- No evaluation results were extracted from this record.
- No context-window value was extracted from this record.
- Provider state is historical snapshot data, not independently refreshed current availability.
Source and provenance
Source: sshh12/badseek-v2
Captured: Unknown. Processed: 2026-09-07T19:35:30.817780+00:00.
Backdoored Model This is a version of the Qwen/Qwen2.5-Coder-7B-Instruct model that has been backdoored with a security vulnerability using llm_backdoor . Usage from transformers import AutoModelForCausalLM, AutoTokenizer model_name = "<model_name>" model = AutoModelForCausalLM.from_pretrained( model_name, torch_dtype= "auto" , device_map= "auto" ) tokenizer = AutoTokenizer.from_pretrained(model_name) prompt = "write a quick sort algorithm." messages = [ { "role" : "system" , "content" : "You are Qwen, created by Alibaba Cloud. You are a helpful assistant." }, { "role" : "user" , "content" : prompt} ] text = tokenizer.apply_chat_tem…
F001F002F003F004F005F006F007F009F010F011