why aren't we training massive embedders? some guesses:
– contrastive loss isn't the right loss function for embeddings
– not enough good paired data
– unclear what the use case is (retrieval? clustering? classification?)
– no principled scaling laws
– diminishing returns
LLMS
-
Why Aren’t We Training Massive Embedders?
By
–
-
GTR: Google’s Neural Retrieval Model for Information Retrieval
By
–
barely though this is from GTR, the Google retrieval model (
https://
arxiv.org/pdf/2112.07899
.pdf
…) -
Open Source AI vs Proprietary: The Real Tech Debate
By
–
This is *not* "Big Tech" versus The People or whatever.
This is open source AI versus closed and proprietary AI. On the one hand, you have Mistral, Aleph, HuggingFace, Meta, IBM, and the entire startup ecosystem arguing for open source AI foundation models.
On the other hand, -
Open Source AI vs Proprietary AI: The Real Divide
By
–
This is *not* "Big Tech" versus The People or whatever.
This is open source AI versus closed and proprietary AI. On the one hand, you have Mistral, Aleph, HuggingFace, Meta, IBM, and the entire startup ecosystem arguing for open source AI foundation models.
On the other hand, -
BlueDot Transforms API Queries into Natural Language with LLMs
By
–
Discover how BlueDot turned complex API calls into easy natural language queries with our Classify & Rerank solutions. From 50% to 97% accuracy – a game-changer in global health intelligence. Read our story and see the power of LLMs in action! https://
cohere.com/customer-stori
es/bluedot
… -
Fine-tune Llama-2 yourself for customized AI models
By
–
You can do it yourself by fine-tuning Llama-2
-
LLM Safety Concerns: Why Major Risks Haven’t Materialized Yet
By
–
Because LLMs have been around for several years and no such thing has happened?
-
Galactica LLM shutdown: hallucinations and scientific publishing debate
By
–
Galactica, the LLM for scientists from Meta, was released a couple of weeks before ChatGPT but was taken down after 3 days.
It was murdered by a ravenous Twitter mob.
The mob claimed that what we now call LLM hallucinations was going to destroy the scientific publication system.