docs(README): update readme

This commit is contained in:
OlivierDehaene 2023-07-25 19:45:25 +02:00
parent a0d55358d2
commit e64a65891b
2 changed files with 38 additions and 39 deletions

View File

@ -42,20 +42,11 @@ python-client-tests:
python-tests: python-server-tests python-client-tests python-tests: python-server-tests python-client-tests
run-bloom-560m: run-falcon-7b-instruct:
text-generation-launcher --model-id bigscience/bloom-560m --num-shard 2 --port 8080 text-generation-launcher --model-id tiiuae/falcon-7b-instruct --port 8080
run-bloom-560m-quantize: run-falcon-7b-instruct-quantize:
text-generation-launcher --model-id bigscience/bloom-560m --num-shard 2 --quantize --port 8080 text-generation-launcher --model-id tiiuae/falcon-7b-instruct --quantize bitsandbytes --port 8080
download-bloom:
HF_HUB_ENABLE_HF_TRANSFER=1 text-generation-server download-weights bigscience/bloom
run-bloom:
text-generation-launcher --model-id bigscience/bloom --num-shard 8 --port 8080
run-bloom-quantize:
text-generation-launcher --model-id bigscience/bloom --num-shard 8 --quantize --port 8080
clean: clean:
rm -rf target aml rm -rf target aml

View File

@ -25,12 +25,12 @@ to power LLMs api-inference widgets.
- [Get Started](#get-started) - [Get Started](#get-started)
- [Docker](#docker) - [Docker](#docker)
- [API Documentation](#api-documentation) - [API Documentation](#api-documentation)
- [Using a private or gated model](#using-a-private-or-gated-model)
- [A note on Shared Memory](#a-note-on-shared-memory-shm) - [A note on Shared Memory](#a-note-on-shared-memory-shm)
- [Distributed Tracing](#distributed-tracing) - [Distributed Tracing](#distributed-tracing)
- [Local Install](#local-install) - [Local Install](#local-install)
- [CUDA Kernels](#cuda-kernels) - [CUDA Kernels](#cuda-kernels)
- [Run BLOOM](#run-bloom) - [Run Falcon](#run-falcon)
- [Download](#download)
- [Run](#run) - [Run](#run)
- [Quantization](#quantization) - [Quantization](#quantization)
- [Develop](#develop) - [Develop](#develop)
@ -81,11 +81,10 @@ or
The easiest way of getting started is using the official Docker container: The easiest way of getting started is using the official Docker container:
```shell ```shell
model=bigscience/bloom-560m model=tiiuae/falcon-7b-instruct
num_shard=2
volume=$PWD/data # share a volume with the Docker container to avoid downloading weights every run volume=$PWD/data # share a volume with the Docker container to avoid downloading weights every run
docker run --gpus all --shm-size 1g -p 8080:80 -v $volume:/data ghcr.io/huggingface/text-generation-inference:0.9 --model-id $model --num-shard $num_shard docker run --gpus all --shm-size 1g -p 8080:80 -v $volume:/data ghcr.io/huggingface/text-generation-inference:0.9.3 --model-id $model
``` ```
**Note:** To use GPUs, you need to install the [NVIDIA Container Toolkit](https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/install-guide.html). We also recommend using NVIDIA drivers with CUDA version 11.8 or higher. **Note:** To use GPUs, you need to install the [NVIDIA Container Toolkit](https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/install-guide.html). We also recommend using NVIDIA drivers with CUDA version 11.8 or higher.
@ -99,14 +98,14 @@ You can then query the model using either the `/generate` or `/generate_stream`
```shell ```shell
curl 127.0.0.1:8080/generate \ curl 127.0.0.1:8080/generate \
-X POST \ -X POST \
-d '{"inputs":"What is Deep Learning?","parameters":{"max_new_tokens":17}}' \ -d '{"inputs":"What is Deep Learning?","parameters":{"max_new_tokens":20}}' \
-H 'Content-Type: application/json' -H 'Content-Type: application/json'
``` ```
```shell ```shell
curl 127.0.0.1:8080/generate_stream \ curl 127.0.0.1:8080/generate_stream \
-X POST \ -X POST \
-d '{"inputs":"What is Deep Learning?","parameters":{"max_new_tokens":17}}' \ -d '{"inputs":"What is Deep Learning?","parameters":{"max_new_tokens":20}}' \
-H 'Content-Type: application/json' -H 'Content-Type: application/json'
``` ```
@ -120,10 +119,10 @@ pip install text-generation
from text_generation import Client from text_generation import Client
client = Client("http://127.0.0.1:8080") client = Client("http://127.0.0.1:8080")
print(client.generate("What is Deep Learning?", max_new_tokens=17).generated_text) print(client.generate("What is Deep Learning?", max_new_tokens=20).generated_text)
text = "" text = ""
for response in client.generate_stream("What is Deep Learning?", max_new_tokens=17): for response in client.generate_stream("What is Deep Learning?", max_new_tokens=20):
if not response.token.special: if not response.token.special:
text += response.token.text text += response.token.text
print(text) print(text)
@ -134,14 +133,26 @@ print(text)
You can consult the OpenAPI documentation of the `text-generation-inference` REST API using the `/docs` route. You can consult the OpenAPI documentation of the `text-generation-inference` REST API using the `/docs` route.
The Swagger UI is also available at: [https://huggingface.github.io/text-generation-inference](https://huggingface.github.io/text-generation-inference). The Swagger UI is also available at: [https://huggingface.github.io/text-generation-inference](https://huggingface.github.io/text-generation-inference).
### Using on private models or gated models ### Using a private or gated model
You can use `HUGGING_FACE_HUB_TOKEN` environment variable to set the token used by `text-generation-inference` to give access to protected ressources. You have the option to utilize the `HUGGING_FACE_HUB_TOKEN` environment variable for configuring the token employed by
`text-generation-inference`. This allows you to gain access to protected resources.
### Distributed Tracing For example, if you want to serve the gated Llama V2 model variants:
`text-generation-inference` is instrumented with distributed tracing using OpenTelemetry. You can use this feature 1. Go to https://huggingface.co/settings/tokens
by setting the address to an OTLP collector with the `--otlp-endpoint` argument. 2. Copy your cli READ token
3. Export `HUGGING_FACE_HUB_TOKEN=<your cli READ token>`
or with Docker:
```shell
model=meta-llama/Llama-2-7b-chat-hf
volume=$PWD/data # share a volume with the Docker container to avoid downloading weights every run
token=<your cli READ token>
docker run --gpus all --shm-size 1g -e HUGGING_FACE_HUB_TOKEN=$token -p 8080:80 -v $volume:/data ghcr.io/huggingface/text-generation-inference:0.9.3 --model-id $model
```
### A note on Shared Memory (shm) ### A note on Shared Memory (shm)
@ -169,6 +180,11 @@ and mounting it to `/dev/shm`.
Finally, you can also disable SHM sharing by using the `NCCL_SHM_DISABLE=1` environment variable. However, note that Finally, you can also disable SHM sharing by using the `NCCL_SHM_DISABLE=1` environment variable. However, note that
this will impact performance. this will impact performance.
### Distributed Tracing
`text-generation-inference` is instrumented with distributed tracing using OpenTelemetry. You can use this feature
by setting the address to an OTLP collector with the `--otlp-endpoint` argument.
### Local install ### Local install
You can also opt to install `text-generation-inference` locally. You can also opt to install `text-generation-inference` locally.
@ -205,7 +221,7 @@ Then run:
```shell ```shell
BUILD_EXTENSIONS=True make install # Install repository and HF/transformer fork with CUDA kernels BUILD_EXTENSIONS=True make install # Install repository and HF/transformer fork with CUDA kernels
make run-bloom-560m make run-falcon-7b-instruct
``` ```
**Note:** on some machines, you may also need the OpenSSL libraries and gcc. On Linux machines, run: **Note:** on some machines, you may also need the OpenSSL libraries and gcc. On Linux machines, run:
@ -221,20 +237,12 @@ the kernels by using the `DISABLE_CUSTOM_KERNELS=True` environment variable.
Be aware that the official Docker image has them enabled by default. Be aware that the official Docker image has them enabled by default.
## Run BLOOM ## Run Falcon
### Download
It is advised to download the weights ahead of time with the following command:
```shell
make download-bloom
```
### Run ### Run
```shell ```shell
make run-bloom # Requires 8xA100 80GB make run-falcon-7b-instruct
``` ```
### Quantization ### Quantization
@ -242,7 +250,7 @@ make run-bloom # Requires 8xA100 80GB
You can also quantize the weights with bitsandbytes to reduce the VRAM requirement: You can also quantize the weights with bitsandbytes to reduce the VRAM requirement:
```shell ```shell
make run-bloom-quantize # Requires 8xA100 40GB make run-falcon-7b-instruct-quantize
``` ```
## Develop ## Develop