Views
No views yet
qwen-bangla-crime-investigation-Q4_K_M.gguf: standalone merged Q4_K_M GGUF for local inference.Qwen/Qwen3.5-9Bjc-builds/Qwen3.5-9B-Q4_K_M-GGUF / Qwen3.5-9B-Q4_K_M.gguftanziro/qwen-bangla-crime-investigation-loratanziro/qwen-bangla-crime-investigation-lora-gguf / qwen-bangla-crime-investigation-lora.F16.gguftanziro/bangla-crime-investigation-patterns-v2qwen-bangla-crime-investigation-Q4_K_M.gguf in LM Studio as a local GGUF model. This repo is different from the LoRA-adapter GGUF repo: this file is merged and does not require loading a separate adapter.1You are a Bangladesh-focused crime investigation and investigative journalism assistant.
2Answer with the final response only. Do not reveal hidden reasoning or step-by-step thinking.
3Separate confirmed facts, source claims, and inference.
4Do not identify private people from anonymized records.
5Avoid doxxing, harassment, vigilantism, or instructions that enable crime or evidence tampering.
6Recommend lawful, ethical next steps: document review, source corroboration, official records, expert consultation, right of reply, and personal safety.0.389918), while training loss continued to improve through step 700.| Step | Training Loss | Validation Loss |
|---|---|---|
| 100 | 0.385574 | 0.409264 |
| 200 | 0.354652 | 0.394317 |
| 300 | 0.356278 | 0.389918 |
| 400 | 0.350144 | 0.391207 |
| 500 | 0.335047 | 0.391896 |
| 600 | 0.318425 | 0.399930 |
| 700 | 0.294148 | 0.406011 |
Thinking Process. Use a system prompt that asks for final answers only. Future fine-tuning should include answer-only examples to suppress chain-of-thought style output.merged GGUF verified and ready2026-05-26 23:49:49 UTCllama-export-lora from a Q4_K_M base GGUF, then llama-quantize --allow-requantize Q4_K_M.