by Ciro Santilli (@cirosantilli, 37)
Homepage: www.llama.com/
Ollama is a highly automated open source wrapper that makes it very easy to run multiple Open weight LLM models either on CPU or GPU.
Its README alone is of great value, serving as a fantastic list of the most popular Open weight LLM models in existence.
Install with:
curl https://ollama.ai/install.sh | sh
The below was tested on Ollama 0.1.14 from December 2013.
Download llama2 7B and open a prompt:
ollama run llama2
On P14s it runs on CPU and generates a few tokens per second, which is quite usable for a quick interactive play.
As mentioned at github.com/jmorganca/ollama/blob/0174665d0e7dcdd8c60390ab2dd07155ef84eb3f/docs/faq.md the downloads to under /usr/share/ollama/.ollama/models/ and ncdu tells me:
--- /usr/share/ollama ----------------------------------
    3.6 GiB [###########################] /.ollama
    4.0 KiB [                           ]  .bashrc
    4.0 KiB [                           ]  .profile
    4.0 KiB [                           ]  .bash_logout
The file:
/usr/share/ollama/.ollama/models/manifests/hf.co/mlabonne/Meta-Llama-3.1-8B-Instruct-abliterated-GGUF/Q2_K
gives a the exact model name and parameters.
We can also do it non-interactively with:
/bin/time ollama run llama2 'What is quantum field theory?'
which gave me:
0.13user 0.17system 2:06.32elapsed 0%CPU (0avgtext+0avgdata 17280maxresident)k
0inputs+0outputs (0major+2203minor)pagefaults 0swaps
but note that there is a random seed that affects each run by default. ollama-expect is an attempt to make the output deterministic.
Some other quick benchmarks from Amazon EC2 GPU on a g4nd.xlarge instance which had an Nvidia Tesla T4:
0.07user 0.05system 0:16.91elapsed 0%CPU (0avgtext+0avgdata 16896maxresident)k
0inputs+0outputs (0major+1960minor)pagefaults 0swaps
and on Nvidia A10G in an g5.xlarge instance:
0.03user 0.05system 0:09.59elapsed 0%CPU (0avgtext+0avgdata 17312maxresident)k
8inputs+0outputs (1major+1934minor)pagefaults 0swaps
So it's not too bad, a small article in 10s.
It tends to babble quite a lot by default, but eventually decides to stop.
This appears to be the backend library of Ollama.
They have a CLI front-end named llama-cli.
askubuntu.com/questions/1461564/install-llama-cpp-locally has some tutorials for Ubuntu. There was no nicely pre-packaged one for Ubuntu 25.04, but build worked on 79e0b68c178656bb0632cb8602d2940b755077f8 In particular it exposed Vulkan support before Ollama did: github.com/ollama/ollama/pull/5059 and it did seem to work, using up my AMD GPU.
A CLI front-end for llama.cpp.
A decent test command:
time ./llama-cli \
  --no-display-prompt \
  --single-turn \
  --temp 0 \
  -c 16384 \
  -cnv \
  -m Llama-3.1-Tulu-3-8B-Q8_0.gguf \
  -n 1000 \
  -ngl 100 \
  -p 'What is quantum field theory?' \
  -t 10 |
  tee output.txt \
;
but it failed to be deterministic despite --temperature 0. This ran 2x faster at 18 tokens/s for 1000 tokens on P14s on GPU via Vulkan than on CPU which is achievable by removing the -ngl 100.

Tagged

It might be easier to just use llama-cli for this, it has a --temperature flag.
Attempt at: ollama-expect
Usage:
./ollama-expect <model> <prompt>
e.g.:
./ollama-expect llama3.2 'What is quantum field theory?'
This generates 100 tokens for the given prompt with the given model.
Benchmarks:

Ancestors (13)

  1. Large language model
  2. Text-to-text model
  3. AI text generation
  4. Generative AI by modality
  5. Generative AI
  6. AI by capability
  7. Artificial intelligence
  8. Machine learning
  9. Computer
  10. Information technology
  11. Area of technology
  12. Technology
  13. Home

View article source

Discussion (0)

+ New discussion

There are no discussions about this article yet.

Articles by others on the same topic (0)

There are currently no matching articles.
See all articles in the same topic + Create my own version