Support llama3.1:70b on compatible systems

[divbell]

I really like this tool and was curious if you could add llama3.1:70b where the system has >40GB of VRAM.

I'm running dual 3090s and can execute it, though it's slow compared to the smaller models, and I would like to benchmark it in a standardized way. I did see one result for this model on the results page though it's unclear to me how they ran it.

Thank you!

[chuangtc]

Not so many people have such powerful machine as you do.
Please checkout the code, then modify the following lines to 70b

https://github.com/aidatatools/ollama-benchmark/blob/c6a5fded472370f35a158258e79f06d36313ffb1/llm_benchmark/data/benchmark1.yml#L17C15-L17C16

ollama-benchmark/llm_benchmark/data/benchmark_models_8gb_ram.yml

Line 9 in c6a5fde

- model: llama3.1:8b

Then run it in development mode. Thank you.