Based on DeepSeek R1. Is it Better?

Описание: Omg the story without e’s. Has three e’s. Not sure how i missed them.


Deep dive into OpenThinker's new 32B and 7B parameter models! Watch as we put these Qwen-based models through a comprehensive test suite to evaluate their reasoning capabilities, response times, and overall performance.

🎯 In this video, you'll learn:
• How OpenThinker models were created using DeepSeek R1 and the OpenThoughts dataset
• Detailed performance analysis of the 32B model across 12 diverse challenges
• Comparison between different quantizations (Q4_K_M, Q8, FP16)
• Why the 7B model shows surprising performance differences

⏱️ Timestamps:
00:00 - Start
01:41 - The Goal of this video
02:16 - Time's impact on the Score
03:26 - First question
04:51 - Second Question
06:03 - Third Question
06:51 - Fourth Question
08:06 - Fifth Question
09:01 - Sixth Question
09:43 - Seventh Question
10:22 - Eighth Question
11:01 - Ninth Question
11:37 - Tenth Question
12:14 - Eleventh Question
12:48 - Twelfth Question
13:43 - The other quants
14:19 - Model Config

🔗 Resources mentioned:
• Ollama.com - Download OpenThinker models

💭 Share your thoughts! Have you tried OpenThinker or DeepSeek? What do you think about using one LLM to train another?

#AI #MachineLearning #OpenThinker #LLM



My Links 🔗
👉🏻 Subscribe (free):    / technovangelist  
👉🏻 Join and Support:    / @technovangelist  
👉🏻 Newsletter: https://technovangelist.substack.com/...
👉🏻 Twitter:   / technovangelist  
👉🏻 Discord:   / discord  
👉🏻 Patreon:   / technovangelist  
👉🏻 Instagram:   / technovangelist  
👉🏻 Threads: https://www.threads.net/@technovangel...
👉🏻 LinkedIn:   / technovangelist  
👉🏻 All Source Code: https://github.com/technovangelist/vi...

Want to sponsor this channel? Let me know what your plans are here: https://www.technovangelist.com/sponsor