My_Unbiased_Opinion

• How much VRAM needed for Qwen 3.6 27B Q8 with 262K context?

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 73 comments

• Qwen3.6 35B A3B Heretic (KLD 0.0015!) Incredible model. Best 35B I have found!

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 145 comments

• Overclocked M40 24GB vs P40 (Benchmark Results)

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 64 comments

• Preserve thinking on or off? (Qwen 3.6)

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 21 comments

• Qwen 3.6 27B: IQ3XXS KV Q8 vs Q4XL KV Q4 (262K context)

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 36 comments

• Open WebUI Desktop Released!

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 111 comments

• Qwen 3.6 27B (IQ3XXS) vs 35B A3B (IQ4XS)?

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 22 comments

• Uncensored huihui-ai/QwQ-32B-abliterated is very good!

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 36 comments

• JOSIEFIED Qwen3 8B is amazing! Uncensored, Useful, and great personality.

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 123 comments

• I think I made the best general use System Prompt for Qwen 3.5 (OpenWebUI + Web search)

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 7 comments

• Trouble with Qwen 3.5 with LMstudio..

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 14 comments

• Qwen-3.5-27B-Derestricted

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 87 comments

• Qwen 3.5 122B A10B - 35.84 score on NatInt (UGI Benchmark)

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 10 comments

• What is currently the best Uncensored LLM for 24gb of VRAM?

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 68 comments

• GLM 4.7 Flash 30B PRISM + Web Search: Very solid.

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 64 comments

• FYI, looks like Tesla P40s are back down in price!

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 46 comments

• Magistral 1.2 is incredible. Wife prefers it over Gemini 2.5 Pro.

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 116 comments

• Anyone try Prime Intellect-3 Prism?

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 1 comments

• Crazy idea: Derestricted Llama 405B?

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 0 comments

• Trying to move to ik_llama.cpp from ollama. Need some help with args..

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 9 comments

• Found an uncensored Qwen2.5 32B!

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 93 comments

• Help: looking for a fast web search option with chat

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 4 comments

• 2100$ PNY 5090 Epic-X OC vs 2000$ 5090FE

  Posted by My_Unbiased_Opinion@reddit | buildapc | View on Reddit | 9 comments

• Abliterated Mistral v0.3 7B?

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 16 comments

• Tesla P40s have doubled in price on eBay.

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 88 comments

• Qwen 3 32B + 8B have less censorship under RAG than other Qwen 3 models.

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 15 comments

• Looking for software that processes images in realtime (or periodically).

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 5 comments

• Help: Gemma 3 High CPU usage during prompt processing?

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 4 comments

• Considering a 9950X for a CPU only Qwen 3 30B A3B..

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 36 comments

• Llama 3.1 8B Instruct abliterated GGUF!

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 62 comments

• Deepseek R1 Distill: 70B iQ2S+imatrix vs 32B iQ4+imatrix

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 9 comments

• Just replaced Llama 3.1 70B @ iQ2S for Qwen 2.5 32B @ Q4KM

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 91 comments

• Will the tariffs decrease the viability of next gen?

  Posted by My_Unbiased_Opinion@reddit | hardware | View on Reddit | 44 comments

• Tesla P40 is too expensive. Here is the next best value. (45$ per 10GB, P102)

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 40 comments

• I think we should train LLMs in increasing complexity while avoiding material on the internet.

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 17 comments

• Cheapest OpenAI API provider of Mistral Large 2?

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 10 comments

• It's been a while since Meta updated Llama..

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 3 comments

• 405B P40 performance?

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 18 comments

• PARAMETER setting question (Ollama/OpenWebUI)

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 1 comments

• My Goofy Ass Inference Server

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 41 comments

• 3090 24gb vs V620 32gb

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 1 comments

• Phi Medium Abliterated: LLM rambles when responding.

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 0 comments

• Anyone willing to Abliterate Hermes 2 Pro Llama 3 8B?

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 0 comments

• Ablitered Mistral 7B v0.3?

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 0 comments

• Getting 5.3 t/s with 70B and a P40 @ IQ2S 4k context. Anyone else get the same?

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 11 comments

• Do multiple GPUs (P40) increase inference speed?

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 47 comments

• Abliterated Gemma2 27B or Qwen2 72B?

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 6 comments

• Will the P40 support Bitnet?

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 15 comments

• Are IQ quants slow on the P40?

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 32 comments

• IQ2XXS @ 8K vs IQ2XS @ 4K vs IQ2S @ 2K?

  Posted by My_Unbiased_Opinion@reddit | LocalLLaMA | View on Reddit | 13 comments