deepseek-chat#

Context Length: 4096
Model Name: deepseek-chat
Languages: en, zh
Abilities: chat
Description: DeepSeek LLM is an advanced language model comprising 67 billion parameters. It has been trained from scratch on a vast dataset of 2 trillion tokens in both English and Chinese.

Specifications#

Model Spec 1 (pytorch, 7 Billion)#

Model Format: pytorch
Model Size (in billions): 7
Quantizations: 4-bit, 8-bit, none
Model ID: deepseek-ai/deepseek-llm-7b-chat
Model Hubs: Hugging Face, ModelScope

Execute the following command to launch the model, remember to replace ${quantization} with your chosen quantization method from the options listed above:

xinference launch --model-name deepseek-chat --size-in-billions 7 --model-format pytorch --quantization ${quantization}

Model Spec 2 (pytorch, 67 Billion)#

Model Format: pytorch
Model Size (in billions): 67
Quantizations: 4-bit, 8-bit, none
Model ID: deepseek-ai/deepseek-llm-67b-chat
Model Hubs: Hugging Face, ModelScope

Execute the following command to launch the model, remember to replace ${quantization} with your chosen quantization method from the options listed above:

xinference launch --model-name deepseek-chat --size-in-billions 67 --model-format pytorch --quantization ${quantization}

Model Spec 3 (ggufv2, 7 Billion)#

Model Format: ggufv2
Model Size (in billions): 7
Quantizations: Q2_K, Q3_K_L, Q3_K_M, Q3_K_S, Q4_0, Q4_K_M, Q4_K_S, Q5_0, Q5_K_M, Q5_K_S, Q6_K, Q8_0
Model ID: TheBloke/deepseek-llm-7B-chat-GGUF
Model Hubs: Hugging Face

Execute the following command to launch the model, remember to replace ${quantization} with your chosen quantization method from the options listed above:

xinference launch --model-name deepseek-chat --size-in-billions 7 --model-format ggufv2 --quantization ${quantization}

Model Spec 4 (ggufv2, 67 Billion)#

Model Format: ggufv2
Model Size (in billions): 67
Quantizations: Q2_K, Q3_K_L, Q3_K_M, Q3_K_S, Q4_0, Q4_K_M, Q4_K_S, Q5_0, Q5_K_M, Q5_K_S, Q6_K, Q8_0
Model ID: TheBloke/deepseek-llm-67b-chat-GGUF
Model Hubs: Hugging Face

Execute the following command to launch the model, remember to replace ${quantization} with your chosen quantization method from the options listed above:

xinference launch --model-name deepseek-chat --size-in-billions 67 --model-format ggufv2 --quantization ${quantization}