LMFlow

LMFlow copied to clipboard

Added a new evaluation metric called ROUGE-L (https://github.com/yizhongw/self-instruct). To apply: run_evaluation_with_rougel.sh Test case: rougel_test_case.sh

YourSaDady

script for prompt tuning

1

**Is your feature request related to a problem? Please describe.** I notice there are script for lora base finetune and evaluation, but not for prompt tuning. **Describe the solution you'd...

xiaoyunwu

[FEATURES] Support image encoder with image caption as example

2

Support image encoder with image caption as example. Try model: BLIP with Salesforce/blip-image-captioning-bas Discussion: + the name of arch_type: visionEncoder_decoder + format the data with image_text + Should we generate...

lianqing11

[BUG] exits with return code = 1 , AttributeError: 'DeepSpeedCPUAdam' object has no attribute 'ds_opt_adam'(Runing on a Linux Server with 6 A100, without Sudo power)

5

... Loading extension module cpu_adam... Traceback (most recent call last): File "/home/mahongli/LMFlow/examples/finetune.py", line 61, in main() File "/home/mahongli/LMFlow/examples/finetune.py", line 57, in main tuned_model = finetuner.tune(model=model, dataset=dataset) File "/home/mahongli/LMFlow/src/lmflow/pipeline/finetuner.py", line 285,...

tonymhl

`preprocessing_num_workers` can not use in `scripts/run_finetune.sh`

6

**Describe the bug** tokenizer map in `hf_decoder_model` use multi `preprocessing_num_workers` will return `TypeError: cannot pickle 'torch._C._distributed_c10d.ProcessGroup' object` **To Reproduce** Steps to reproduce the behavior: add `--preprocessing_num_workers 20 \` to `scripts/run_finetune.sh`...

csyourui

CUDA_VISIBLE_DEVICES=0 \ deepspeed examples/evaluate.py \ --answer_type text \ --model_name_or_path output_models/llama-7b-hf \ --lora_model_path output_models/instruction_ckpt/llama7b-lora \ --dataset_path data/alpaca/test \ --prompt_structure "Input: {input}" \ --deepspeed examples/ds_config.json ~

2

i user scripts/run_evaluation_with_lora.sh CUDA_VISIBLE_DEVICES=0 \ deepspeed examples/evaluate.py \ --answer_type text \ --model_name_or_path output_models/llama-7b-hf \ --lora_model_path output_models/instruction_ckpt/llama7b-lora \ --dataset_path data/alpaca/test \ --prompt_structure "Input: {input}" \ --deepspeed examples/ds_config.json ~ then result 2023-06-10...

dadaxxxx

Is it possible to evaluate while training

2

我看到 run_finetune_with_lora_save_aggregated_weights.sh中 有以下参数 --do_train \ --do_eval \ --evaluation_strategy "steps" \ --eval_steps 1000 \ --eval_dataset_path ${eval_dataset_path} \ 是否可以 修改 run_finetune_with_lora.sh 加入上述参数，从而实现一边训练一边评估 我尝试了下，抛以下错误：  能否解惑，谢谢！

abc20220327

How to use pycharm single-step debugging . Tips: run remote in server

3

I edit the configurations of finetune.py in pycharm as below: ======================================= --model_name_or_path facebook/galactica-1.3b --dataset_path /root/LMFlow/data/alpaca/train --output_dir /root/LMFlow/output_models/finetune_with_lora --overwrite_output_dir --num_train_epochs 0.01 --learning_rate 1e-4 --block_size 512 --per_device_train_batch_size 1 --use_lora 1 --lora_r 8...

huizhilei

RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cpu and cuda:0! (when checking argument for argument index in method wrapper_CUDA_gather)

1

when you meet this problem, you have to check your transformers first. Further, you may need to change the following code: file: transformers/src/transformers/models/llama/modeling_llama.py **change** ``` def apply_rotary_pos_emb(q, k, cos, sin,...

seanxuu

[Error][launch.py:324:sigkill_handler]exits with return -7

3

作者好，我的报错情况如下图所示，我的显卡是3090，使用的是docker容器运行，就是从docker hub上拉取的镜像。 

Agreewithu