Hacker News: therealoliver

DeepDive: In-Depth Decryption of LLMs Construction and Inference from Scratch

therealoliver — Wed, 26 Feb 2025 13:46:04 +0000

Article URL: https://github.com/therealoliver/Deepdive-llama3-from-scratch/blob/main/README.md

Comments URL: https://news.ycombinator.com/item?id=43183558

Points: 2

# Comments: 0

New comment by therealoliver in "DeepDive in everything of Llama3: revealing detailed insights and implementation"

therealoliver — Sat, 22 Feb 2025 16:49:29 +0000

Thanks for ur feedback on the project's visual style. Let me clarify a few things: First, associating an anime character, especially a minor, with inappropriate content is unfair and disrespectful to both the character and the original work. Second, this character wasn’t added by me—it was included by the previous developer who created these visualizations. As part of the open-source community, I believe it’s important to respect the project’s history and acknowledge their contributions. Lastly, I appreciate your perspective—it might highlight how cultural differences can shape perceptions, and I’ll pay attention to that in my subsequent work. That said, I also value maintaining the entertainment and avoiding overly dry or preachy content, which I think is equally important. This will likely be my final reply on this matter. Let’s focus on being open-minded and keeping the community peaceful. Thanks again for your support of the technical aspects of the project

New comment by therealoliver in "DeepDive in everything of Llama3: revealing detailed insights and implementation"

therealoliver — Sat, 22 Feb 2025 02:17:59 +0000

Thank you for helping me out. Such comments are really depressing.

New comment by therealoliver in "DeepDive in everything of Llama3: revealing detailed insights and implementation"

therealoliver — Sat, 22 Feb 2025 02:12:54 +0000

glad to have helped you :)

New comment by therealoliver in "DeepDive in everything of Llama3: revealing detailed insights and implementation"

therealoliver — Sat, 22 Feb 2025 02:11:52 +0000

I'm glad to have helped you :)

New comment by therealoliver in "DeepDive in everything of Llama3: revealing detailed insights and implementation"

therealoliver — Sat, 22 Feb 2025 02:10:23 +0000

Yes, these are two different learning paths. The detailed process learning is beneficial for future research, while the API-style approach is convenient and quick for getting started and using. Both are very useful!

DeepDive in everything of Llama3: revealing detailed insights and implementation