Great video comparison of Mixtral12B vs. Llama 3.2 90B Vision! It really highlights the strengths and limitations of each model. It's surprising to see how the smaller 12B handled the table formatting better, while Llama 3.2 90B struggled but showed promising improvements in data extraction accuracy. GPT-4 Omni's performance was also a nice addition. Exciting times for AI advancements! 🔥💡
1.5 so bad… for commercial use case when you need to imagination purpose and helping in daily context use open ai, for code no better than anthropic claude
Just uploaded a post that the performance numbers by the community indicate, that Qwen2-VL-72b-Instruct is better than LLama 3.2 90B Vision. Score of 1095 to 1076.
Yes, I continue a good additional hour to talk to Llama 3.2 90B to increase their performance, because I thought that a 90B model should realize a higher pattern complexity than a 11B model, but I failed. therefore I just cut it. smile.
Great video comparison of Mixtral12B vs. Llama 3.2 90B Vision! It really highlights the strengths and limitations of each model. It's surprising to see how the smaller 12B handled the table formatting better, while Llama 3.2 90B struggled but showed promising improvements in data extraction accuracy. GPT-4 Omni's performance was also a nice addition. Exciting times for AI advancements! 🔥💡
Pixtral vs Qwen72-VL vs Molmo 72B => Which one is the best?
Is your tool open source? I want to compare models after fine tuning
What tool do you use to make the comparison ?
lmarena.ai/
Curious how this compares to Gemini 1.5.
1.5 so bad… for commercial use case when you need to imagination purpose and helping in daily context use open ai, for code no better than anthropic claude
Have you tried with Qwen2-vl
Just uploaded a post that the performance numbers by the community indicate, that Qwen2-VL-72b-Instruct is better than LLama 3.2 90B Vision. Score of 1095 to 1076.
What UI do you use?
looks like llmsys evaluation website
@@antoninleroy3863 tank's
Quite the abrupt end 😅
Yes, I continue a good additional hour to talk to Llama 3.2 90B to increase their performance, because I thought that a 90B model should realize a higher pattern complexity than a 11B model, but I failed. therefore I just cut it. smile.