Inference support for T5 and FLAN-T5 model families #8141
fairydreaming
merged
18 commits into
ggml-org:master
from
fairydreaming:t5-clean-3
llama : add inference support and model types for T5 and FLAN-T5 mode…
45681a57
Merge branch 'ggerganov:master' into t5-clean-3
1c8d37a2
llama : updated llm_build_ffn() calls to new API in build_t5()
bad0cafe
llama : make pos_bias contiguous for CUDA
c4ded1a8
Merge remote-tracking branch 'upstream/master' into t5-clean-3
7293243d
llama : whitespace formatting
7d7fff46
vladfaust
commented
on 2024-06-28
llama : quantization-related fixes for T5
6dc9eb40
ggerganov
commented
on 2024-07-02
llama : add early return in Unigram tokenizer when normalized input i…
78675f35
llama : remove obsolete code
1d1cb01b
add t5 tokenizer tests
7c610faf
Merge remote-tracking branch 'upstream/master' into t5-clean-3
b01ce7df
llama : move JAIS after T5 everywhere for easier merging later
d40c9a1d
llama : change naming to prefer "_enc" suffix
03ab5dd6
llama : simplify llama_encode_internal
88270a36
llama-batched : add encoder support
ded682d4
llama : minor
01cd5a66
ggerganov
approved these changes
on 2024-07-04
llama : silence compiler warnings
8b560e63
Merge branch 'ggerganov:master' into t5-clean-3
9bcecf1d
Assignees
Labels
examples
python
Review Complexity : Medium
Milestone
No milestone
Login to write a write a comment.
Login via GitHub