PLBjt

9 ms·

1.
by PLBjt 1y ago
Is the CEO technical or not?
2.
by PLBjt 2y ago
Hi, depends on what kind of model you want to fine-tune. Could you be a little more specific?
3.
by PLBjt 2y ago
Hi HN! I'm PLB from phospho. I’m building a dev kit for ML engineers to work with real-world robots using bearable dev tools (Python, HTTP APIs, WebSocket…) Some Python examples: basic robot controls ( https://docs.phospho.ai
4.
Show HN: Practice YC interviews with an AI partner using OpenAI's real-time API(github.com)
1 points by PLBjt 2y ago|0 comments
5.
Train a linear layer to increase your RAG system [Technical Report] [pdf](research.phospho.ai)
3 points by PLBjt 2y ago|1 comments
6.
by PLBjt 2y ago
Key highlights: - Uses a simple linear transformation on existing embeddings - Boosts hit rate from 89% to 95% on real-world examples - Minor increase on latency (less than 10ms) - Works on top of blackbox embedding models (Mistral AI, Open
7.
Embedding user intention in text messages as dense vectors [pdf](research.phospho.ai)
4 points by PLBjt 2y ago|1 comments
8.
Uncensored LLM vision API (for internal use cases)(docs.phospho.ai)
1 points by PLBjt 2y ago|1 comments
9.
by PLBjt 2y ago
The safety filter of GPT 4 vision and Claude 3 Opus is critical for user facing products, but to build internal tools it impacts a lot performance (eg. to build an image moderation LLM). We had this problem internally and we had to build ou
10.
by PLBjt 3y ago
Portkey is a proxy, so you will have to call their API, and they will route the request to a LLM provider. Then, they provide you with some observability tools to look at these requests -> LLM router + observability phospho is not a prox