jphme/em_german_leo_mistral

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:7BQuant:FP8Ctx Length:8kPublished:Oct 7, 2023License:apache-2.0Architecture:Transformer0.1K Open Weights Warm

jphme/em_german_leo_mistral is a 7 billion parameter instruction-tuned causal language model developed by jphme, based on the LeoLM Mistral architecture with a context length of 8192 tokens. This model is specifically optimized for German language proficiency, having undergone continued pretraining with German texts. It excels at understanding, generating, and interacting with German content, making it suitable for German-centric NLP applications.

Loading preview...

What is EM German Leo Mistral?

jphme/em_german_leo_mistral is a 7 billion parameter instruction-tuned model from the EM German family, built upon the LeoLM Mistral architecture. It stands out as a highly proficient open-source German LLM, benefiting from continued pretraining on extensive German text datasets by the LeoLM team. This model is designed to offer superior performance in German language tasks compared to other models in its class, balancing strong capabilities with reasonable computational requirements.

Key Capabilities

  • Optimized for German Language: Fine-tuned on a large dataset of German instructions, ensuring high proficiency in understanding and generating German text.
  • LeoLM Base: Utilizes a LeoLM base model, which is a Llama-2/Mistral variant with enhanced German pretraining.
  • Instruction Following: Capable of following instructions in German, making it suitable for various conversational and generative tasks.
  • Efficient Performance: Recommended as the best combination of performance and computing requirements within the EM German model family.

Should I use this for my use case?

  • German-centric Applications: Ideal for any application requiring high-quality German text generation, comprehension, or interaction.
  • Resource-Conscious Deployment: A strong choice when seeking robust German language capabilities without the computational overhead of larger models.
  • Instruction-Following Tasks: Well-suited for tasks that involve responding to specific instructions or prompts in German.
  • Research and Development: Recommended for research purposes, particularly in German NLP, with a note to consider the base model's license.

Popular Sampler Settings

Top 3 parameter combinations used by Featherless users for this model. Click a tab to see each config.

temperature
top_p
top_k
frequency_penalty
presence_penalty
repetition_penalty
min_p