netikas

• New open weights models: GigaChat-3.1-Ultra-702B and GigaChat-3.1-Lightning-10B-A1.8B

  Posted by netikas@reddit | LocalLLaMA | View on Reddit | 176 comments

• Actual comparison between locally ran Qwen-3.6-27B and proprietary models

  Posted by netikas@reddit | LocalLLaMA | View on Reddit | 72 comments

• Entropy-Adaptive Finetuning

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

• Which opensource models are best for (kinda) RP in French and German?

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

• What’s up with vllm?

  Posted by netikas@reddit | LocalLLaMA | View on Reddit | 8 comments

• Why bother with RWKV/Mamba instead of decoder transformers?

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

• Qwen-2-72B-Instruct in HF Chat is kinda broken?

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

• Aya-23-8b for multilingual RAG

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