I took a random book excerpt with 23,000 words (±30k input tokens) and used it as context. Jev still responds in 800ms, sometimes 500ms. That's in the neighbourhood of 20-50,000 tok/s prefill, which is obviously not possible with normal LLMs, not even Cerebras is this fast.
The prefill is the only blocking part, and you can prefill the whole context up to the point where they diverge, then prefill each question and decode the one token in parallel for each question.
If you batch vLLM calls with the same prompt prefix to the same process, it'll deduplicate the prompt prefix across batched requests (+/- the block size) and decode in parallel for each.
That's with a vanilla LLM. If you modify the LLM you can pull that in-graph, but it isn't really necessary.
ricardobeat · · focus · HN ↗
I took a random book excerpt with 23,000 words (±30k input tokens) and used it as context. Jev still responds in 800ms, sometimes 500ms. That's in the neighbourhood of 20-50,000 tok/s prefill, which is obviously not possible with normal LLMs, not even Cerebras is this fast.
stingraycharles · · focus · HN ↗
Xorlev · · focus · HN ↗
The prefill is the only blocking part, and you can prefill the whole context up to the point where they diverge, then prefill each question and decode the one token in parallel for each question.
If you batch vLLM calls with the same prompt prefix to the same process, it'll deduplicate the prompt prefix across batched requests (+/- the block size) and decode in parallel for each.
That's with a vanilla LLM. If you modify the LLM you can pull that in-graph, but it isn't really necessary.
paul_irolla · · focus · HN ↗
[dead]