Skip to content

Commit

Permalink
Merge branch 'google:main' into tokenizer
Browse files Browse the repository at this point in the history
  • Loading branch information
bhavya01 authored Apr 25, 2024
2 parents 807e4e9 + 9469054 commit 67f9c00
Show file tree
Hide file tree
Showing 2 changed files with 7 additions and 4 deletions.
2 changes: 1 addition & 1 deletion README.md
Original file line number Diff line number Diff line change
Expand Up @@ -16,7 +16,7 @@ Currently, there are two reference engine implementations available -- one for J
### Jax

- Git: https://github.com/google/maxtext
- README: https://github.com/google/JetStream/blob/main/jetstream/max_text/README.md
- README: https://github.com/google/JetStream/blob/main/docs/online-inference-with-maxtext-engine.md

### Pytorch

Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -126,7 +126,7 @@ export ICI_AUTOREGRESSIVE_PARALLELISM=-1
export ICI_TENSOR_PARALLELISM=1
export SCAN_LAYERS=false
export WEIGHT_DTYPE=bfloat16
export PER_DEVICE_BATCH_SIZE=6
export PER_DEVICE_BATCH_SIZE=4
```

#### Create Llama2-13b environment variables for server flags
Expand Down Expand Up @@ -191,7 +191,10 @@ Note: these flags are from [MaxText config](https://github.com/google/maxtext/bl

```bash
cd ~
python JetStream/jetstream/tools/requester.py
# For Gemma model
python JetStream/jetstream/tools/requester.py --tokenizer maxtext/assets/tokenizer.gemma
# For Llama2 model
python JetStream/jetstream/tools/requester.py --tokenizer maxtext/assets/tokenizer.llama2
```

The output will be similar to the following:
Expand Down Expand Up @@ -250,7 +253,7 @@ wget https://huggingface.co/datasets/anon8231489123/ShareGPT_Vicuna_unfiltered/r
# run benchmark with the downloaded dataset and the tokenizer in maxtext
# You can control the qps by setting `--request-rate`, the default value is inf.
python JetStream/benchmarks/benchmark_serving.py \
--tokenizer /home/$USER/maxtext/assets/tokenizer.gemma \
--tokenizer maxtext/assets/tokenizer.gemma \
--num-prompts 1000 \
--dataset sharegpt \
--dataset-path ~/ShareGPT_V3_unfiltered_cleaned_split.json \
Expand Down

0 comments on commit 67f9c00

Please sign in to comment.