]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
model : add support for talkie-1930-13b (#22596)
authorNiklas Sheth <redacted>
Tue, 26 May 2026 04:57:38 +0000 (00:57 -0400)
committerGitHub <redacted>
Tue, 26 May 2026 04:57:38 +0000 (07:57 +0300)
commitc9d98295a3a371c5a9277b54be493ec74317b387
tree2bc98fc8010f1c12e152180096d97a5286c39cde
parent1506d39e76d1a6694a3a9412ed3116feb7b37ac1
model : add support for talkie-1930-13b (#22596)

* initial talkie support, coherent

* reorder to follow convention

* absorb inverse rope

* stop folding scalars to improve quantization

* use broadcasting instead of duplication

* style cleanup

* add scaling support to LoraTorchTensor; use that path in conversion

* use layer_out_scale instead of embd_skip_scale
14 files changed:
conversion/__init__.py
conversion/base.py
conversion/talkie.py [new file with mode: 0644]
convert_hf_to_gguf_update.py
convert_lora_to_gguf.py
gguf-py/gguf/constants.py
gguf-py/gguf/tensor_mapping.py
src/llama-arch.cpp
src/llama-arch.h
src/llama-model.cpp
src/llama-model.h
src/llama-vocab.cpp
src/models/models.h
src/models/talkie.cpp [new file with mode: 0644]