danielmiessler / fabric

fabric is an open-source framework for augmenting humans using AI. It provides a modular framework for solving specific problems using a crowdsourced set of AI prompts that can be used anywhere.
https://danielmiessler.com/p/fabric-origin-story
MIT License
24.74k stars 2.64k forks source link

[Question]: I have ollama installed (locally and remotely) - still stuck - probably some dependence on API_KEY #448

Closed fmotta closed 2 months ago

fmotta commented 5 months ago

What is your question?

[Question]: I have ollama installed (locally and remotely) -

After reading the documentation, I am still not clear how to get fabric working.

I tried this, this: pbpaste | fabric --stream --remoteOllamaServer $OLLAMA_HOST --model llava --pattern analyze_claims

And I got: Error: 'NoneType' object has no attribute 'chat' 'NoneType' object has no attribute 'chat'

echo $OLLAMA_HOST 192.168.1.178

~$ ollama list NAME ID SIZE MODIFIED llama2-uncensored:latest 44040b922233 3.8 GB 6 hours ago llama3:latest 365c0bd3c000 4.7 GB 2 days ago llava:latest 8dd30f6b0cb1 4.7 GB 3 days ago dolphin-mistral:latest 5dc8c5a2be65 4.1 GB 3 days ago nomic-embed-text:latest 0a109f422b47 274 MB 3 days ago

Windows 10 WSL Ubuntu 20.04

How do I get past this ?

snafu4 commented 5 months ago

Try

pbpaste | fabric --stream --model llava:latest --pattern analyze_claims

On May 29, 2024 at 5:52 PM -0400, danielmiessler/fabric @.***>, wrote:

pbpaste | fabric --stream --remoteOllamaServer $OLLAMA_HOST --model llava --pattern analyze_claims

fmotta commented 5 months ago

pbpaste | fabric --stream --model llava:latest --pattern analyze_claims

Thanks! This seemed to intuit my use of ollama it seems. :)

It is interesting to see the difference in results by changing llava with llama2-uncensiored.

Oddly llava did far more and seemed to abide by the pattern and llama2-uncensored just informs about the story (cut fragment from hitchhikers guide wiki)

snafu4 commented 5 months ago

Np.

Not surprisingly, the different models are ‘good’ at different things.

Other than the models with suffixes like ‘code’ or ‘Python’, I wonder if there is a website somewhere that explicitly states what a given model is good (and not good) at providing information on. On May 29, 2024 at 6:43 PM -0400, fmotta @.***>, wrote:

pbpaste | fabric --stream --model llava:latest --pattern analyze_claims Thanks! This seemed to intuit my use of ollama it seems. :) It is interesting to see the difference in results by changing llava with llama2-uncensiored. Oddly llava did far more and seemed to abide by the pattern and llama2-uncensored just informs about the story (cut fragment from hitchhikers guide wiki) — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.***>