huggingface/text-generation-inference
5
0
Fork 0
mirror of https://github.com/huggingface/text-generation-inference.git synced 2025-09-11 12:24:53 +00:00
Code Issues Packages Projects Releases Wiki Activity

update doc

Browse Source
This commit is contained in:
OlivierDehaene 2024-01-26 16:27:27 +01:00
parent 7d2bc40c42
commit 4ee87f41ab
1 changed files with 10 additions and 18 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

28
docs/source/basic_tutorials/launcher.md
View File

@ -60,13 +60,21 @@ Options:
[env: QUANTIZE=] [env: QUANTIZE=]
Possible values: Possible values:
- awq: 4 bit quantization. Requires a specific GTPQ quantized model: https://hf.co/models?search=awq. Should replace GPTQ models whereever possible because of the better latency - awq: 4 bit quantization. Requires a specific AWQ quantized model: https://hf.co/models?search=awq. Should replace GPTQ models wherever possible because of the better latency
- eetq: 8 bit quantization, doesn't require specific model. Should be a drop-in replacement to bitsandbytes with much better performance. Kernels are from https://github.com/NetEase-FuXi/EETQ.git - eetq: 8 bit quantization, doesn't require specific model. Should be a drop-in replacement to bitsandbytes with much better performance. Kernels are from https://github.com/NetEase-FuXi/EETQ.git
- gptq: 4 bit quantization. Requires a specific GTPQ quantized model: https://hf.co/models?search=gptq. text-generation-inference will use exllama (faster) kernels whereever possible, and use triton kernel (wider support) when it's not. AWQ has faster kernels - gptq: 4 bit quantization. Requires a specific GTPQ quantized model: https://hf.co/models?search=gptq. text-generation-inference will use exllama (faster) kernels wherever possible, and use triton kernel (wider support) when it's not. AWQ has faster kernels
- bitsandbytes: Bitsandbytes 8bit. Can be applied on any model, will cut the memory requirement in half, but it is known that the model will be much slower to run than the native f16 - bitsandbytes: Bitsandbytes 8bit. Can be applied on any model, will cut the memory requirement in half, but it is known that the model will be much slower to run than the native f16
- bitsandbytes-nf4: Bitsandbytes 4bit. Can be applied on any model, will cut the memory requirement by 4x, but it is known that the model will be much slower to run than the native f16 - bitsandbytes-nf4: Bitsandbytes 4bit. Can be applied on any model, will cut the memory requirement by 4x, but it is known that the model will be much slower to run than the native f16
- bitsandbytes-fp4: Bitsandbytes 4bit. nf4 should be preferred in most cases but maybe this one has better perplexity performance for you model - bitsandbytes-fp4: Bitsandbytes 4bit. nf4 should be preferred in most cases but maybe this one has better perplexity performance for you model
```
## SPECULATE
```shell
--speculate <SPECULATE>
The number of input_ids to speculate on If using a medusa model, the heads will be picked up automatically Other wise, it will use n-gram speculation which is relatively free in terms of compute, but the speedup heavily depends on the task
[env: SPECULATE=]
``` ```
## DTYPE ## DTYPE
```shell ```shell
@ -290,22 +298,6 @@ Options:
[env: ROPE_FACTOR=] [env: ROPE_FACTOR=]
```
## SLIDING_WINDOW
```shell
--sliding-window <SLIDING_WINDOW>
Sliding Window will only be used by flash attention optimized models Limit the Paged Attention context window size
[env: SLIDING_WINDOW=]
```
## ATTENTION_SINKS
```shell
--attention-sinks <ATTENTION_SINKS>
If `sliding_window` is set, always keep the first `attention_sinks` tokens in the context See: [Efficient Streaming Language Models with Attention Sinks](https://arxiv.org/abs/2309.17453)
[env: ATTENTION_SINKS=]
``` ```
## JSON_OUTPUT ## JSON_OUTPUT
```shell ```shell