WetHat💦
@wethat.fosstodon.org.ap.brid.gy
I enjoy programming computers.
#fedi22 #Mastodon #Fediverse #Programming #OpenSource #FOSS #MachineLearning #ArtificialIntelligence #AI #Python #CommonLisp […]
[bridged from https://fosstodon.org/@WetHat on the fediverse by https://fed.brid.gy/ ]
#fedi22 #Mastodon #Fediverse #Programming #OpenSource #FOSS #MachineLearning #ArtificialIntelligence #AI #Python #CommonLisp […]
[bridged from https://fosstodon.org/@WetHat on the fediverse by https://fed.brid.gy/ ]
Here is part 2 of the article with more examples of critical failures:
https://hackernoon.com/hallucinations-by-design-part-2-the-silent-flaws-of-embeddings-and-why-your-ai-is-getting-it-wrong?source=rss
https://hackernoon.com/hallucinations-by-design-part-2-the-silent-flaws-of-embeddings-and-why-your-ai-is-getting-it-wrong?source=rss
Hallucinations by Design (Part 2): The Silent Flaws of Embeddings & Why Your AI Is Getting It Wrong | HackerNoon
Embedding and LLM's needs to be tested and evaluated or hallucinations will happen. Experimentation and evaluation on custom data is a must - openai and genai
hackernoon.com
April 2, 2025 at 7:55 AM
Here is part 2 of the article with more examples of critical failures:
https://hackernoon.com/hallucinations-by-design-part-2-the-silent-flaws-of-embeddings-and-why-your-ai-is-getting-it-wrong?source=rss
https://hackernoon.com/hallucinations-by-design-part-2-the-silent-flaws-of-embeddings-and-why-your-ai-is-getting-it-wrong?source=rss
The article is a bit short on explaining the quality metric:
The ' Cosine similarity' measures the cosine of the angle between two embedding vectors, representing two pieces of text. The value ranges from -1 to 1, where:
- **1** indicates the texts are very similar.
- **0** indicates that the […]
The ' Cosine similarity' measures the cosine of the angle between two embedding vectors, representing two pieces of text. The value ranges from -1 to 1, where:
- **1** indicates the texts are very similar.
- **0** indicates that the […]
Original post on fosstodon.org
fosstodon.org
April 2, 2025 at 7:40 AM
The article is a bit short on explaining the quality metric:
The ' Cosine similarity' measures the cosine of the angle between two embedding vectors, representing two pieces of text. The value ranges from -1 to 1, where:
- **1** indicates the texts are very similar.
- **0** indicates that the […]
The ' Cosine similarity' measures the cosine of the angle between two embedding vectors, representing two pieces of text. The value ranges from -1 to 1, where:
- **1** indicates the texts are very similar.
- **0** indicates that the […]