oxeo0/opsec-blogposts
1
0
Fork 0
mirror of http://git.nowherejezfoltodf4jiyl6r56jnzintap5vyjlia7fkirfsnfizflqd.onion/nihilist/opsec-blogposts.git synced 2025-05-16 19:27:06 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

fix tables

Browse source
This commit is contained in:
oxeo0 2025-05-07 11:07:55 +02:00
parent cd713e7e44
commit 24e5443f34
3 changed files with 13 additions and 24 deletions
Download patch file Download diff file Expand all files Collapse all files

8
openwebuilocalllms/index.md
View file

@ -67,8 +67,10 @@ We'll show how to check prompt length and set appropriate context size in Open W
**Open LLMs - Past and Present**
Ever since [LLaMA 1 was leaked](https://www.theverge.com/2023/3/8/23629362/meta-ai-language-model-llama-leak-online-misuse) on 4chan in February 2023, we started seeing more and more companies caring about so called "open" LLMs. Those can be downloaded and run on users computer without any restrictions.
In early 2025 we also heard A LOT about deepseek-r1. An open reasoning LLM that tried to compete with OpenAI's ChatGPT. While it wasn't Deepseeks first model (their earlier work includes deepseek-coder, deepseek-v2.5 MoE and deepseek-v3), the hype was global. It brought a lot of attention to open LLM ecosystem and made proprietary AI companies loose billions.
Personally, I was interested in Open LLMs since their inception - when ollama project based on llama.cpp was born. Here're some of my picks for best self-hostable models currently available on ollama: | Gemma 3 | Phi4 | Granite 3.2* | Qwen 2.5 | llama 3.1
---|---|---|---|---|---
Personally, I was interested in Open LLMs since their inception - when ollama project based on llama.cpp was born. Here're some of my picks for best self-hostable models currently available on ollama:
- | Gemma 3 | Phi4 | Granite 3.2* | Qwen 2.5 | llama 3.1
| ---|---|---|---|---|---
Trained by | Google | Microsoft | IBM | Alibaba | Meta
Released on | 03-2025 | 12-2024 | 02-2025 | 09-2024 | 07-2024
Parameter count | 12B | 14B | 8B | 14B | 8B
@ -77,7 +79,7 @@ Other variants | 1B, 4B, 27B | Phi4-mini 4B | Vision 2B, 2B | 0.5B, 1.5B, 3B, 7B
Good at | Reasoning, explaining concepts, light programming, vision, translation | Instruction following, programming, complex tasks, translation | General tasks, light programming, translation | Translation, general tasks | General tasks
Bad at | - | - | Complex tasks | Instruction following | Complex tasks, instruction following
* - while I was writing this tutorial, Granite 3.3 was released
\* - while I was writing this tutorial, Granite 3.3 was released
## **Use-Cases**