Kimi - K2 - Thinking 3.825bit MLX is a quantized model for text generation. It achieves different perplexity performances in tests through different quantization methods. Among them, q3.825bit quantization can reach a perplexity of 1.256.

Natural Language ProcessingMlxEnglish

Kimi K2 Thinking is the latest generation of open-source thinking model developed by Moonshot AI. It has powerful deep reasoning ability and tool call function. The model adopts a mixture-of-experts architecture, supports native INT4 quantization, has a 256k context window, and performs excellently in multiple benchmark tests.

Natural Language ProcessingTransformers

This is a high-performance 4-bit quantized version of the Kimi K2 Instruct model, optimized for running on Apple Silicon (M1/M2/M3/M4) Macs using the MLX framework. The model has 671 billion parameters, supports a 128K context window, and achieves an excellent balance between quality and efficiency, making it an ideal choice for most practical deployments.

Natural Language ProcessingMlxMultiple Languages

This is the quantized version of the LLM360/K2-Think model, dedicated to making knowledge spread more freely, adhering to the concept of 'Let knowledge be used by everyone'. This version optimizes the model size and inference efficiency through quantization technology.

Natural Language ProcessingGguf

Kimi - K2 - Instruct is a quantized version based on the moonshotai/Kimi - K2 - Instruct model, aiming to make knowledge more widely accessible to the public.

Natural Language ProcessingGguf

Kimi K2 is an advanced Mixture of Experts (MoE) language model with 32 billion active parameters and 1 trillion total parameters, optimized for agent capabilities.

Natural Language ProcessingTransformers

Kimi K2 is an advanced Mixture of Experts (MoE) language model with 32 billion active parameters and 1 trillion total parameters, optimized for intelligent agent capabilities.

Natural Language ProcessingTransformers

Kimi K2 is an advanced Mixture of Experts (MoE) language model with 32 billion active parameters and a total of 1 trillion parameters, optimized for agent capabilities.

Natural Language ProcessingTransformers

Kimi K2 is an advanced Mixture of Experts (MoE) language model with 32 billion active parameters and 1 trillion total parameters, optimized for agent capabilities.

Natural Language ProcessingTransformers

K2 is a large language model with 65 billion parameters, surpassing Llama 2 70B with a 35% computational advantage through a fully transparent training approach.

Natural Language ProcessingTransformersEnglish

Korean large language model fine-tuned based on SOLAR-10.7B-v1.0, specializing in Korean understanding and generation tasks

Natural Language ProcessingTransformersKorean

Icefall is an automatic speech recognition (ASR) toolkit based on the k2 framework, focusing on efficient and flexible speech recognition model training and inference.

A T5-based model capable of converting input keywords into coherent sentences.

Natural Language ProcessingJaxEnglish

A T5-based model capable of converting input keywords into complete sentences.

Natural Language ProcessingTransformersEnglish