Ampere® optimized Ollama.
1.2K
Ampere® optimized build of Ollama.
This Docker image can be run on bare metal Ampere® CPUs and Ampere® based VMs available in the cloud.
For best results we recommend using models in our custom quantization formats available on Ampere's HuggingFace page.
This image is a drop-in replacement for the original Ollama image. It includes performance improvements while maintaining full compatibility with the original's features and behavior.
Start the server:
docker run --privileged=true --name ollama -p 11434:11434 amperecomputingai/ollama:latest
In a separate shell:
docker exec -it ollama bash
ollama run hf.co/AmpereComputing/llama-3.2-3b-instruct-gguf:Llama-3.2-3B-Instruct-Q8R16.gguf
or using the REST API:
curl http://localhost:11434/api/pull -d '{
"model": "hf.co/AmpereComputing/llama-3.2-3b-instruct-gguf:Llama-3.2-3B-Instruct-Q8R16.gguf"
}'
curl http://localhost:11434/api/generate -d '{
"model": "hf.co/AmpereComputing/llama-3.2-3b-instruct-gguf:Llama-3.2-3B-Instruct-Q8R16.gguf",
"prompt":"Why is the sky blue?"
}'
Ampere® optimized Ollama provides support for two new quantization methods, Q4_K_4 and Q8R16, offering model size and perplexity similar to Q4_K and Q8_0, respectively, but performing up to 1.5-2x faster on inference. You can download them from Ampere's HuggingFace page.
You can also quantize the model yourself using Ampere® optimized llama.cpp and following the instructions.
Afterwards, you can use it with Ollama by creating a Modelfile:
FROM ./path/to/your/model.gguf
and running
ollama create my_model -f Modelfile
ollama run my_model
Please contact us at [email protected]
By accessing, downloading or using this software and any required dependent software (the “Ampere AI Software”), you agree to the terms and conditions of the software license agreements for the Ampere AI Software, which may also include notices, disclaimers, or license terms for third party software included with the Ampere AI Software. Please refer to the Ampere AI Software EULA v1.6 or other similarly-named text file for additional details.
Content type
Image
Digest
sha256:c2cde79ea…
Size
91.2 MB
Last updated
about 1 year ago
docker pull amperecomputingai/ollama