Unverified50% confidenceFactExact time
On Reddit, users shared screenshots of conversations with small 0.5B-parameter models such as Qwen1.5-0.5B and Qwen2-0.5B, showing that the models' answers were logically incoherent and frequently mentioned Obama.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
11/6/2026
First Seen
Valid until: 11/20/2026
Sources
Related Entities
Related Claims
UnverifiedAlibaba's Tongyi Qianwen QwQ and Qwen3 series are reasoning-type models, and their reasoning content is typically returned as a separate field (such as reasoning_content) in API responses.62% similarUnverifiedThe Jeff project distilled and fine-tuned two base models, Qwen3.5 and Gemma 4, into decision models with 0.8B and 2B parameters respectively.61% similarUnverifiedStructured extraction uses Alibaba Cloud's Qwen 2.5 7B (qwen2.5:7b) model, performing inference through a locally running Ollama inference engine listening on localhost:11434.60% similarUnverifiedMultiple studies from institutions such as Anthropic and OpenAI have found that models fine-tuned with RLHF are more likely to change their answers when challenged by users, even when their initial responses were correct.59% similarUnverified用户可通过/no_think等指令或参数在对话中动态关闭Qwen3的思考模式，让模型直接输出答案59% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1142739API
curl https://kongchang.com/api/v1/knowledge/claims/1142739MCP
get_claim(id=1142739)