llama.cpp
model : add support for MiniMaxText01ForCausalLM and MiniMaxM1ForCausalLM
#27018
Merged

model : add support for MiniMaxText01ForCausalLM and MiniMaxM1ForCausalLM #27018

fairydreaming
sszymczy llama : support for MiniMax-Text-01 model
d9bee6ec
sszymczy Merge remote-tracking branch 'upstream/master'
5bdd054f
sszymczy chore : renames to match the other MiniMax models
bd605684
sszymczy model : add logits mask as MiniMax-Text-01 embeddings tensor has zero…
9a641467
sszymczy llama : replace hardcoded conditions with hparams.is_recr()
09b462e0
sszymczy model : used build_rs() for recurrent state management
5ad0114c
sszymczy chore : code cleanup
fc81d9e5
sszymczy model : optimized MiniMax-Text-01 by removing the state tranpose oper…
0cc3e784
sszymczy chore : removed unnecessary ggml_cont() in MiniMax-Text-01 implementa…
9eae5976
sszymczy llama : add generic logits mask graph input
8256d01f
sszymczy model : permuted diag_decay dimensions to avoid doing it inside MiniM…
3cc40e6f
sszymczy chore : code cleanup
4e750d7d
sszymczy Merge remote-tracking branch 'upstream/master' into minimax-text-01
d73091b1
sszymczy chore : code cleanup
e0f434d9
sszymczy model : use token positions when calculating MiniMax-Text-01 decay te…
2286b408
sszymczy convert : add support for MiniMaxM1ForCausalLM as it seems to be the …
bf3be2d1
sszymczy chat : add jinja template for MiniMax-M1
414b9fc4
sszymczy chore : code cleanup
e5a2a3b4
fairydreaming fairydreaming requested a review from CISC CISC 8 days ago
fairydreaming fairydreaming requested a review from ggerganov ggerganov 8 days ago
github-actions github-actions added model
github-actions github-actions added conversion
sszymczy tests : MINIMAX_01-related fixes
24297134
sszymczy chore : silence Python lint errors
ac55c3ee
fairydreaming fairydreaming requested a review from JohannesGaessler JohannesGaessler 7 days ago
github-actions github-actions added testing
CISC
CISC commented on 2026-08-13
sszymczy vocab : remove unnecessary vocab type
f708076e
sszymczy convert : update MiniMaxText01Model conversion to use yield when modi…
58c32e81
ngxson
ngxson commented on 2026-08-13
sszymczy convert : suppress tokens with zero-valued embeddings during MiniMax-…
a07a0169
sszymczy llama : removed logits mask - no longer necessary as token suppressio…
84df482e
fairydreaming fairydreaming force pushed from 3c36eeb0 to 84df482e 7 days ago
CISC
CISC approved these changes on 2026-08-14
fairydreaming model : use common functions to make MiniMax-Text-01 implementation m…
82b4cf1a
fairydreaming model : use common functions to make MiniMax-Text-01 implementation m…
ec9a9fc3
CISC
fairydreaming
CISC
sszymczy convert : override non-working built-in chat template during conversion
b837059a
sszymczy tests : skip arch MINIMAX_01 tests for WebGPU backend (it breaks again)
d6736604
fairydreaming
CISC
CISC approved these changes on 2026-08-14
CISC CISC requested a review from ngxson ngxson 6 days ago
ngxson
CISC
ngxson
ngxson
ngxson approved these changes on 2026-08-14
ngxson ngxson merged 16d222fc into master 6 days ago
yomaytk

Login to write a write a comment.

Login via GitHub

Assignees
No one assigned
Labels
Milestone