Skip to content

Latest commit

 

History

History
125 lines (87 loc) · 4.6 KB

inference.md

File metadata and controls

125 lines (87 loc) · 4.6 KB
Raw

Inference

We demonstrate how to run inference (next token prediction) with the GPT base model in the litgpt generate command:

litgpt generate stabilityai/stablelm-base-alpha-3b \
  --prompt "Hello, my name is"

Output:

Hello, my name is Levi Durrer, I'm an Austrian journalist - Chairman of the Press Blair Party, with 37 years in the Press Blair International, and two years in the Spectre of Austerity for the other. I'm crossing my fingers that you will feel

The script assumes you have downloaded and converted the weights as described here.

This will run the 3B pre-trained model and require ~7 GB of GPU memory using the bfloat16 datatype.

Run interactively

You can also chat with the model interactively:

litgpt chat stabilityai/stablelm-tuned-alpha-3b

This script can work with any checkpoint. For the best chat-like experience, we recommend using it with a checkpoints fine-tuned for chatting such as stabilityai/stablelm-tuned-alpha-3b or togethercomputer/RedPajama-INCITE-Chat-3B-v1.

Tip

Use --multiline true to work with inputs that span multiple lines.

Run a large model on one smaller device

Check out our quantization tutorial.

Run a large model on multiple smaller devices

We offer two scripts to leverage multiple devices for inference.

litgpt generate_sequentially

Allows you to run models that wouldn't fit in a single card by partitioning the transformer blocks across all your devices and running them sequentially.

For instance, meta-llama/Llama-2-70b-chat-hf would require ~140 GB of GPU memory to load on a single device, plus the memory for activations. With 80 transformer layers, we could partition them across 8, 5, 4, or 2 devices.

litgpt generate_sequentially meta-llama/Llama-2-70b-chat-hf \
  --max_new_tokens 256 \
  --num_samples 2

Using A100 40GB GPUs, we need to use at least 4. You can control the number of devices by setting the CUDA_VISIBLE_DEVICES= environment variable.

Devices Max GPU RAM Token/sec
2 OOM -
4 35.64 GB 7.55
5 28.72 GB 7.49
8 18.35 GB 7.47

Note that the memory usage will also depend on the max_new_tokens value used.

The script also supports quantization, using 4-bit precision, we can now use 2 GPUs

litgpt generate_sequentially meta-llama/Llama-2-70b-chat-hf \
  --max_new_tokens 256 \
  --num_samples 2 \
  --quantize bnb.nf4-dq
Devices Max GPU RAM Token/sec
2 20.00 GB 8.63
4 10.80 GB 8.23
5 8.96 GB 8.10
8 6.23 GB 8.18

Smaller devices can also be used to run inference with this technique.

litgpt generate_tp

Uses tensor parallelism (TP) to run models that wouldn't fit in a single card by sharding the MLP and Attention QKV linear layers across all your devices.

For instance, meta-llama/Llama-2-70b-chat-hf would require ~140 GB of GPU memory to load on a single device, plus the memory for activations. The requirement is that the intermediate size (for the MLP) and the QKV size (for attention) is divisible by the number of devices. With an intermediate size of 28672, we can use 2, 4, 7, or 8 devices. With a QKV size of 10240 we can use 2, 4, 5, or 8 devices. Since the script is configured to shard both, the intersection is used: we can only use 2, 4, or 8 devices.

litgpt generate_tp meta-llama/Llama-2-70b-chat-hf \
  --max_new_tokens 256 \
  --num_samples 2

Using A100 40GB GPUs, we need to use at least 4. You can control the number of devices by setting the CUDA_VISIBLE_DEVICES= environment variable.

Devices Max GPU RAM Token/sec
2 OOM -
4 35.46 GB 9.33
8 18.19 GB 8.61

Note that the memory usage will also depend on the max_new_tokens value used.

The script also supports quantization, using 4-bit precision, we can now use 2 GPUs

litgpt generate_tp meta-llama/Llama-2-70b-chat-hf \
  --max_new_tokens 256 \
  --num_samples 2 \
  --quantize bnb.nf4-dq
Devices Max GPU RAM Token/sec
2 19.79 GB 6.72
4 10.73 GB 6.48
8 6.15 GB 6.20

Smaller devices can also be used to run inference with this technique.