llama.cpp
llama : simplify Mamba with advanced batch splits
#8526
Merged

llama : simplify Mamba with advanced batch splits #8526

compilade merged 19 commits into master from compilade/batch-splits
compilade
compilade llama : advanced batch splits
c51daefc
compilade Merge branch 'master' into compilade/batch-splits
22504ec6
github-actions github-actions added ggml
compilade compilade added refactoring
compilade compilade added Review Complexity : Medium
compilade compilade marked this pull request as draft 1 year ago
compilade llama : fix integer signedness mixing
2e4adb47
compilade llama : logits_all has priority over batch->logits
7b7db0bb
ggerganov
ggerganov commented on 2024-07-17
github-actions github-actions added testing
ggerganov ggerganov force pushed to 7b7db0bb 1 year ago
compilade llama : apply suggestions
1fb5d4fd
compilade llama : fix t5 segfault
1725de76
ggerganov
compilade
hackey
compilade Merge branch 'master' into compilade/batch-splits
9c0a61f8
compilade
hackey
ggerganov
compilade Merge branch 'master' into compilade/batch-splits
0dea4263
compilade llama : fix Mamba session save and restore
704a3033
awgr
compilade llama : minor cosmetic changes
952ed35b
compilade Merge branch 'master' into compilade/batch-splits
5679a3bd
compilade llama : rename llama_reorder_outputs to llama_output_reorder
cfd5a113
compilade compilade marked this pull request as ready for review 1 year ago
ggerganov minor : add struct members for clarity
0596a99f
ggerganov
ggerganov approved these changes on 2024-08-09
compilade Merge branch 'master' into compilade/batch-splits
702e1995
compilade
compilade llama : fix T5 segfault again
652e9b0d
compilade
compilade llama : fix Mamba pooled embeddings with multiple sequences
b264eddb
compilade llama : add llama_model_is_recurrent to simplify figuring that out
1be5ea7d
compilade
compilade compilade added merge ready
compilade Merge branch 'master' into compilade/batch-splits
80d9d2a5
ggerganov
ggerganov approved these changes on 2024-08-21
compilade llama : fix simple splits when the batch contains embeddings
80626503
compilade compilade merged a1631e53 into master 1 year ago
awgr
mann1x

Login to write a write a comment.

Login via GitHub

Reviewers
Assignees
No one assigned
Labels
Milestone