Skip to content

Fix convert script cannot generate bf16 weights #104

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
May 28, 2024
Merged

Fix convert script cannot generate bf16 weights #104

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 2 additions & 1 deletion README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -70,7 +70,8 @@ Need to manually modify the `config.json` in the checkpoint folder to make it a
export input_ckpt_dir=Original llama weights directory
export output_ckpt_dir=The output directory
export model_name="llama-3" # or "llama-2", "gemma"
export quantize_type="int8_per_channel" # Availabe quantize type: {"int8", "int4"} x {"per_channel", "blockwise"}, setting this will quantize the weights
export quantize_weights=True # Whether to quantize weights
export quantize_type="int8_per_channel" # "quantize_weights" needs to be turned on. Availabe quantize type: {"int8", "int4"} x {"per_channel", "blockwise"}, "int8_per_channel" is the default option if not specified.
python -m convert_checkpoints --model_name=$model_name --input_checkpoint_dir=$input_ckpt_dir --output_checkpoint_dir=$output_ckpt_dir --quantize_type=$quantize_type
```

Expand Down
2 changes: 1 addition & 1 deletion convert_checkpoints.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -391,7 +391,7 @@ def main(argv) -> None:
llama_model.Transformer.get_quantized_embedding_weight_to_scaler_map()
)

if FLAGS.quantize_type:
if FLAGS.quantize_weights:
quantize_num_bits = 8 if "int8" in FLAGS.quantize_type else 4
is_blockwise = "blockwise" in FLAGS.quantize_type
weight_axis = lambda x: 0 if x in quantize_embedding_weight_map else 1
Expand Down