Fine-tune of Qwen3-4B cuts AI writing tics, released for Ollama
On 40 held-out prompts the author's slop score fell from 121 to 12.5 per 1,000 words, with a 30-question quiz still correct.
36
AI model
Qwen3-4B-Instruct
- Mentions · 7d1
- Mentions · 30d1
- All mentions2
- First seenSep 20, 2026
- Last mentioned6d ago
Timeline
2 articles · newest firstSimulated students that make realistic mistakes help AI tutors learn faster
Microsoft's StudentSim builds individual student replicas on Qwen3-4B-Instruct to train AI tutors, outperforming GPT-5.4 in chess, English, and math.
40
Appears with
- Ollamaproduct · 1
- qwen3-4b-deslopai model · 1
- Study Modeproduct · 1
- Microsoftvendor · 1
- Maia2ai model · 1
- Alibabavendor · 1
- GPT-5.4ai model · 1
- Microsoftorganization · 1
- Googlevendor · 1
- OpenAIvendor · 1
- Copilotproduct · 1
- StudentSimproduct · 1
- Guided Learningproduct · 1
Entities are extracted by the model from each article. First seen , last mentioned . Watching keeps it in this browser only (no account); the watchlist page and dashboard alerts use it.