Skip to content

Conversation

@ggerganov
Copy link
Member

@ggerganov ggerganov commented Nov 29, 2024

ref #10552

  • Use F16 KV cache for the draft model
  • Set draft context equal to slot context
  • Do not speculate during prompt processing

@ggerganov ggerganov marked this pull request as ready for review December 3, 2024 08:47
@ggerganov ggerganov changed the title server : force F16 KV cache for the draft model server : fix default draft model parameters Dec 3, 2024
@ggerganov ggerganov merged commit 70b98fa into master Dec 3, 2024
43 of 50 checks passed
@ggerganov ggerganov deleted the gg/server-force-draft-kv-f16 branch December 3, 2024 09:20
tinglou pushed a commit to tinglou/llama.cpp that referenced this pull request Dec 7, 2024
* server : force F16 KV cache for the draft model

ggml-ci

* server : fix draft params

ggml-ci

* server : various params fixes

ggml-ci
arthw pushed a commit to arthw/llama.cpp that referenced this pull request Dec 20, 2024
* server : force F16 KV cache for the draft model

ggml-ci

* server : fix draft params

ggml-ci

* server : various params fixes

ggml-ci
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant