diff options
| author | Carolinabanana <140120812+Carolinabanana@users.noreply.github.com> | 2024-04-09 09:16:13 +0100 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2024-04-09 11:16:13 +0300 |
| commit | 5dc9dd7152dedc6046b646855585bd070c91e8c8 (patch) | |
| tree | d2bae3652d91cdd9327e28fa85d167a67e050c53 /convert-hf-to-gguf.py | |
| parent | e11a8999b5690f810c2c99c14347f0834e68c524 (diff) | |
llama : add Command R Plus support (#6491)
* Add Command R Plus GGUF
* Add Command R Plus GGUF
* Loading works up to LayerNorm2D
* Export new tensors in 1D so they are not quantized.
* Fix embedding layer based on Noeda's example
* Whitespace
* Add line
* Fix unexpected tokens on MPS. Re-add F16 fix. ((Noeda)
* dranger003: Fix block index overflow in CUDA dequantizing.
* Reverted blocked multiplication code as it still has issues and could affect other Llama arches
* export norms as f32
* fix overflow issues during quant and other cleanup
* Type convention
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* dranger003: Fix more int overflow during quant.
---------
Co-authored-by: S <seast@Ss-Mac-Studio.local>
Co-authored-by: S <s@example.com>
Co-authored-by: slaren <slarengh@gmail.com>
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
Diffstat (limited to 'convert-hf-to-gguf.py')
| -rwxr-xr-x | convert-hf-to-gguf.py | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/convert-hf-to-gguf.py b/convert-hf-to-gguf.py index 7e601170..37af6328 100755 --- a/convert-hf-to-gguf.py +++ b/convert-hf-to-gguf.py @@ -160,7 +160,7 @@ class Model(ABC): data = data.astype(np.float32) # TODO: Why cant we use these float16 as-is? There should be not reason to store float16 as float32 - if self.ftype == 1 and data_dtype == np.float16 and n_dims == 1: + if self.ftype == 1 and data_dtype == np.float16 and (n_dims == 1 or new_name.endswith("_norm.weight")): data = data.astype(np.float32) # if f16 desired, convert any float32 2-dim weight tensors to float16 |