# German Optimization

A fine-tuned model for German speech recognition based on Whisper Large v3, specifically optimized for German speech processing and recognition.

A German-optimized large language model based on Meta Llama3-8B, continuously pre-trained on 65 billion German tokens, specifically optimized for German and supporting 32k long context

llama3-8b-spaetzle-v20 is a model merged from llama3-8b-spaetzle-v13 and llama-3-wissenschaft-8B-v2, supporting English and German, with strong performance on EQ-Bench v2_de.

DiscoLM German 7b is a large language model based on Mistral, specializing in German applications, and serves as the successor to the EM German model family. This model has undergone SFT fine-tuning on extensive German and English instruction datasets and has been further optimized through DPO reinforcement learning.

DiscoLM 70b is a 70B-parameter multilingual model developed based on Laion's LeoLM 70b, enhanced with additional German text pre-training and instruction fine-tuning to strengthen its English and German capabilities

EM German Leo Mistral is a German large language model based on the Mistral architecture, focusing on German text generation tasks.

EM German is a family of German large language models based on Mistral/LeoLM, fine-tuned on German instruction data, excelling in understanding and generating German text.