Code and documents of LongLoRA and LongAlpaca
[2023.10.18] We support StreamingLLM inference on our LongAlpaca models. This increases the context-length of the multi-round dialogue in StreamingLLM.
[2023.10.8] We release the long instruction-following dataset, LongAlpaca-12k and the corresponding models, LongAlpaca-7B, LongAlpaca-13B, and LongAlpaca-70B.
(The previous sft models, Llama-2-13b-chat-longlora-32k-sft and Llama-2-70b-chat-longlora-32k-sft, have been deprecated.)
[2023.10.3] We add support GPTNeoX models. Please refer to this PR for usage. Thanks for @naubull2 for this contribution.
[2023.9.22] We release all our fine-tuned models, including 70B-32k models, LLaMA2-LongLoRA-70B-32k, LLaMA2-LongLoRA-7B-100k. Welcome to check them out!
[2023.9.22] We release Paper and this GitHub repo, including training and evaluation code.
(Updated Oct 12th 10pm in GMT+8) The demo link is not permanent. We update it every 72h.
LongLoRA: Efficient Fine-tuning of Long-Context Large Language Models [Paper]
Yukang Chen,
Shengju Qian,
Haotian Tang,
Xin Lai,
Zhijian Liu,
Song Han,
Jiaya Jia
Requirements
and Installation and Quick Guide
sections below.To download and use the pre-trained weights you will need:
To install and run the application:
pip install -r requirements.txt
pip install flash-attn --no-build-isolation
LongAlpaca-12k contains 9k long QA data that we collected and 3k short QA sampled from the original Alpaca data. This is to avoid the case that the model might degrade at short instruction following. The data we collect contains various types and amounts as the following figure.
Data | Short QA | Long QA | Total | Download |
---|---|---|---|---|
LongAlpaca-12k | 3k | 9k | 12k | Link |
Following the original Alpaca format, our Long QA data uses the following prompts for fine-tuning:
instruction
: str
, describes the task the model should perform. For example, to answer a question after reading a book section or paper. We vary the contents and questions to make instructions diverse.output
: str
, the answer to the instruction.We did not use the input
format in the Alpaca format for simplicity.
Model | Size | Context | Train | Link |
---|---|---|---|---|
LongAlpaca-7B | 7B | 32768 | Full FT | Model |
LongAlpaca-13B | 13B | 32768 | Full FT | Model |
LongAlpaca-70B | 70B | 32768 | LoRA+ | Model (LoRA-weight) |
Model | Size | Context | Train | Link |
---|---|---|---|---|
Llama-2-7b-longlora-8k-ft | 7B | 8192 | Full FT | Model |
Llama-2-7b-longlora-16k-ft | 7B | 16384 | Full FT | Model |
Llama-2-7b-longlora-32k-ft | 7B | 32768 | Full FT | Model |
Llama-2-7b-longlora-100k-ft | 7B | 100000 | Full FT | Model |
Llama-2-13b-longlora-8k-ft | 13B | 8192 | Full FT | Model |
Llama-2-13b-longlora-16k-ft | 13B | 16384 | Full FT | Model |
Llama-2-13b-longlora-32k-ft | 13B | 32768 | Full FT | Model |
Model | Size | Context | Train | Link |
---|---|---|---|---|
Llama-2-7b-longlora-8k | 7B | 8192 | LoRA+ | LoRA-weight |
Llama-2-7b-longlora-16k | 7B | 16384 | LoRA+ | LoRA-weight |
Llama-2-7b-longlora-32k | 7B | 32768 | LoRA+ | LoRA-weight |
Llama-2-13b-longlora-8k | 13B | 8192 | LoRA+ | LoRA-weight |
Llama-2-13b-longlora-16k | 13B | 16384 | LoRA+ | LoRA-weight |
Llama-2-13b-longlora-32k | 13B | 32768 | LoRA+ | LoRA-weight |
Llama-2-13b-longlora-64k | 13B | 65536 | LoRA+ | LoRA-weight |
Llama-2-70b-longlora-32k | 70B | 32768 | LoRA+ | LoRA-weight |
Llama-2-70b-chat-longlora-32k | 70B | 32768 | LoRA+ | LoRA-weight |
We use LLaMA2 models as the pre-trained weights and fine-tune them to long context window sizes. Download based on your choices.
Pre-trained weights |
---|
Llama-2-7b-hf |
Llama-2-13b-hf |
Llama-2-70b-hf |
Llama-2-7b-chat-hf |
Llama-2-13b-chat-hf |
Llama-2-70b-chat-hf |
This project also supports GPTNeoX models as the base model architecture. Some candidate pre-trained weights may include GPT-NeoX-20B, Polyglot-ko-12.8B and other variants.
torchrun --nproc_per_node=8 fine-tune.py \
--model_name_or_path path_to/Llama-2-7b-hf \
--bf16 True \
--output_dir path_to_saving_checkpoints \
--cache_dir path_to_cache \
--model_max_length 8192 \
--use_flash_attn True \
--low_rank_training False \
--num_train_epochs 1 \
--per_device_train_batch_size 1 \
--per_device_eval_batch_size 2 \
--gradient_accumulation_steps 8 \
--evaluation_strategy "no" \
--save_strategy "steps" \
--save_steps 1000 \
--save_total_limit 2 \
--learning_rate 2e-5 \
--weight_decay 0.0 \
--warmup_steps 20 \
--lr_scheduler_type "constant_with_warmup" \
--logging_steps 1 \
--deepspeed "ds_configs/stage2.json" \
--tf32 True \
--max_steps 1000
path_to/Llama-2-7b-hf
, path_to_saving_checkpoints
, path_to_cache
to your own directory.model_max_length
to other values.ds_configs/stage2.json
to ds_configs/stage3.json
if you want.use_flash_attn
as False
if you use V100 machines or do not install flash attention.low_rank_training
as False
if you want to use fully fine-tuning. It will cost more GPU memory and slower, but the performance will be a bit better.cd path_to_saving_checkpoints && python zero_to_fp32.py . pytorch_model.bin
torchrun --nproc_per_node=8 supervised-fine-tune.py \
--model_name_or_path path_to_Llama2_chat_models \
--bf16 True \
--output_dir path_to_saving_checkpoints \
--model_max_length 32768 \
--use_flash_attn True \
--data_path LongAlpaca-12k.json \
--low_rank_training True \
--num_train_epochs 3 \
--per_device_train_batch_size 1 \
--per_device_eval_batch_size 2 \
--gradient_accumulation_steps 1 \
--evaluation_strategy "no" \
--save_strategy "steps" \
--save_steps 1000 \
--save_total_limit 2 \
--learning_rate 2e-5 \
--weight_decay 0.0 \
--warmup_steps 20 \
--lr_scheduler_type "constant_with_warmup" \
--logging_steps 1 \
--deepspeed "ds_configs/stage2.json" \
--tf32 True
In low-rank training, we set embedding and normalization layers as trainable. Please use the following line to extract the trainable weights trainable_params.bin
from pytorch_model.bin
python3 get_trainable_weights.py --checkpoint_path path_to_saving_checkpoints --trainable_params "embed,norm"
Merge the LoRA weights of pytorch_model.bin
and trainable parameters trainable_params.bin
, save the resulting model into your desired path in the Hugging Face format:
python3 merge_lora_weights_and_save_hf_model.py \
--base_model path_to/Llama-2-7b-hf \
--peft_model path_to_saving_checkpoints \
--context_size 8192 \
--save_path path_to_saving_merged_model
For example,
python3 merge_lora_weights_and_save_hf_model.py \
--base_model /dataset/pretrained-models/Llama-2-7b-hf \
--peft_model /dataset/yukangchen/hf_models/lora-models/Llama-2-7b-longlora-8k \
--context_size 8192 \
--save_path /dataset/yukangchen/models/Llama-2-7b-longlora-8k-merged
To evaluate a model that is trained in the low-rank setting, please set both base_model
and peft_model
. base_model
is the pre-trained weight. peft_model
is the path to the saved checkpoint, which should contain trainable_params.bin
, adapter_model.bin
and adapter_config.json
. For example,
python3 eval.py --seq_len 8192 --context_size 8192 --batch_size 1 --base_model path_to/Llama-2-7b-hf --peft_model path_to_saving_checkpoints --data_path pg19/test.bin
To evaluate a model that is fully fine-tuned, you only need to set base_model
as the path to the saved checkpoint, which should contain pytorch_model.bin
and config.json
. peft_model
should be ignored.
python3 eval.py --seq_len 8192 --context_size 8192 --batch_size 1 --base_model path_to_saving_checkpoints --data_path pg19/test.bin
Note that --seq_len
is to set the sequence length for evaluation. --context_size
is to set the context length of the model during fine-tuning. --seq_len
should not be larger than --context_size
.
We have already tokenized the validation and test splits of PG19 and proof-pile dataset into pg19/validation.bin
, pg19/test.bin
, and proof-pile/test_sampled_data.bin
, with the tokenizer of LLaMA. proof-pile/test_sampled_data.bin
contains 128 documents that are randomly sampled from the total proof-pile test split. For each document, it has at least 32768 tokens. We also release the sampled ids in proof-pile/test_sampled_ids.bin. You can download them from the links below.
Dataset | Split | Link |
---|---|---|
PG19 | validation | pg19/validation.bin |
PG19 | test | pg19/test.bin |
Proof-pile | test | proof-pile/test_sampled_data.bin |
We provide a manner to test the passkey retrieval accuracy. For example,
python3 passkey_retrivial.py \
--context_size 32768 \
--base_model path_to/Llama-2-7b-longlora-32k \
--max_tokens 32768 \
--interval 1000
context_size
is the context length during fine-tuning.max_tokens
is maximum length for the document in passkey retrieval evaluation.interval
is the interval during the document length increasing. It is a rough number because the document increases by sentences.To chat with LongAlpaca models,
python3 inference.py \
--base_model path_to_model \
--question $question \
--context_size $context_length \
--max_gen_len $max_gen_len \
--flash_attn True \
--material $material_content
To ask a question related to a book:
python3 inference.py \
--base_model /data/models/LongAlpaca-13B \
--question "Why doesn't Professor Snape seem to like Harry?" \
--context_size 32768 \
--max_gen_len 512 \
--flash_attn True \
--material "materials/Harry Potter and the Philosophers Stone_section2.txt"
To ask a question related to a paper:
python3 inference.py \
--base_model /data/models/LongAlpaca-13B \
--question "What are the main contributions and novelties of this work?" \
--context_size 32768 \
--max_gen_len 512 \
--flash_attn True \
--material "materials/paper1.txt"
To deploy your own demo run
python3 demo.py \
--base_model path_to_model \
--context_size $context_size \
--max_gen_len $max_gen_len \
--flash_attn True
Example
python3 demo.py \
--base_model /data/models/LongAlpaca-13B \
--context_size 32768 \
--max_gen_len 512 \
--flash_attn True
flash_attn=True
will make the generation slow but save much GPU memory.We support the inference of LongAlpaca models with StreamingLLM. This increases the context-length of the multi-round dialogue in StreamingLLM. Here is an example,
python run_streaming_llama_longalpaca.py \
----enable_streaming \
--test_filepath outputs_stream.json \
--use_flash_attn True \
--recent_size 32768
test_filepath
is the json file that contains prompts for inference. We provide an example file outputs_stream.json, which is a subset of LongAlpaca-12k. You can replace it to your own questions.During our dataset collection, we convert paper and books from pdf to text. The conversion quality has a large influence on the final model quality. We think that this step is non-trivial. We release the tool for the pdf2txt conversion, in the folder pdf2txt
. It is built upon pdf2image
, easyocr
, ditod
and detectron2
. Please refer to the README.md in pdf2txt
for more details.
If you find this project useful in your research, please consider citing:
@article{longlora,
title={LongLoRA: Efficient Fine-tuning of Long-Context Large Language Models},
author={Yukang Chen and Shengju Qian and Haotian Tang and Xin Lai and Zhijian Liu and Song Han and Jiaya Jia},
journal={arXiv:2309.12307},
year={2023}
}
@misc{long-alpaca,
author = {Yukang Chen and Shaozuo Yu and Shengju Qian and Haotian Tang and Xin Lai and Zhijian Liu and Song Han and Jiaya Jia},
title = {Long Alpaca: Long-context Instruction-following models},
year = {2023},
publisher = {GitHub},
journal = {GitHub repository},
howpublished = {\url{https://github.com/dvlab-research/LongLoRA}},
}