(ollama.com)

344 points LorenDB | 2 comments | 16 May 25 01:43 UTC | HN request time: 0.001s | source

Show context

tommica ◴[16 May 25 05:10 UTC] No.44002018[source]▶

Sidetangent: why is ollama frowned upon by some people? I've never really got any other explanation than "you should run llama.CPP yourself"

buyucu ◴[16 May 25 07:20 UTC] No.44002621[source]▶

I abandoned Ollama because Ollama does not support Vulkan: https://news.ycombinator.com/item?id=42886680

You have to support Vulkan if you care about consumer hardware. Ollama devs clearly don't.

replies(1): >>44003156 #

ramon156[dead post] ◴[16 May 25 08:50 UTC] No.44003156[source]▶

[flagged]

buyucu ◴[16 May 25 10:37 UTC] No.44003743[source]▶

why would I use a software that doesn't have the features I want, when a far better alternative like llama.cpp exists? ollama does not add any value.

replies(1): >>44003903 #

1. magicalhippo ◴[16 May 25 11:07 UTC] No.44003903{3}[source]▶

I more often than not add multiple models to my WebUI chats to compare and contrast models.

Ollama makes this trivial compared to llama.cpp, and so for me adds a lot of value due to this.

replies(1): >>44005537 #

2. buyucu ◴[16 May 25 13:51 UTC] No.44005537[source]▶

llama-swap does it better than ollama I think.

Ollama's new engine for multimodal models