Mistral NeMo Mistral NeMo is a new 12B AI model developed with NVIDIA, featuring a 128k token context window that surpasses current benchmarks in reasoning, world knowledge, and coding accuracy for its size. It utilizes a new, more efficient tokenizer called Tekken and is designed for global, multilingual applications, showing particular strength in various languages. Released under the Apache 2.0 license, it’s available as a drop-in replacement for Mistral 7B systems and packaged for NVIDIA NIM inference.

• Mistral NeMo is a 12B parameter AI model created in collaboration with NVIDIA.
• It features a large context window of up to 128k tokens.
• The model demonstrates state-of-the-art performance in reasoning, world knowledge, and coding accuracy for its size.
• It uses a new, more efficient tokenizer named Tekken, which is based on Tiktoken and trained on over 100 languages.
• Mistral NeMo is designed for global, multilingual applications and excels in English, French, German, Spanish, Italian, Portuguese, Chinese, Japanese, Korean, Arabic, and Hindi.
• Pre-trained base and instruction-tuned checkpoints are released under the Apache 2.0 license.
• The model supports FP8 inference without performance loss due to quantisation-aware training.
• Mistral NeMo is available on HuggingFace and as an NVIDIA NIM inference microservice. Continue reading https://foxvector.com/articles/319261cf-6879-4254-9b90-da4a8d9c9c37