[BUG] Data fetch error - typo

[anthisyme]

是否已有关于该错误的issue或讨论？ | Is there an existing issue / discussion for this?

• [X] 我已经搜索过已有的issues和讨论 | I have searched the existing issues / discussions

该问题是否在FAQ中有解答？ | Is there an existing answer for this in FAQ?

• [X] 我已经搜索过FAQ | I have searched FAQ

当前行为 | Current Behavior

Data fetch error raised because of typo on https://github.com/OpenBMB/MiniCPM-V/blob/0e4ec319cf69c6d17b5aa714cbaec29276c84089/finetune/dataset.py#L383 conversation -> conversations

期望行为 | Expected Behavior

No response

复现方法 | Steps To Reproduce

No response

运行环境 | Environment

No response

备注 | Anything else?

No response

[YifanLiang-hust]

the same bug i met QAQ

[LDLINGLINGLING]

hello，can you show me your train data json.

[KeepFaithMe]

在使用finetune_lora.sh测试微调的时候，我也遇到了同样的问题。 麻烦帮忙看一下是哪里出现了问题，谢谢啦！ 这是我的finetune_lora.sh设置 #!/bin/bash

GPUS_PER_NODE=1 NNODES=1 NODE_RANK=0 MASTER_ADDR=localhost MASTER_PORT=6001

MODEL="/work/MiniCPM-V-main/check_point/OpenBMB/MiniCPM-V-2_6-int4"

ATTENTION: specify the path to your training data, which should be a json file consisting of a list of conversations.

See the section for finetuning in README for more information.

DATA="/work/MiniCPM-V-main/minicpm_data/data/train.json" EVAL_DATA="/work/MiniCPM-V-main/minicpm_data/eval/eval.json" LLM_TYPE="minicpm"

if use openbmb/MiniCPM-V-2, please set LLM_TYPE=minicpm

#if use openbmb/MiniCPM-Llama3-V-2_5, please set LLM_TYPE=llama3 export NCCL_P2P_DISABLE=1 # export NCCL_IB_DISABLE=1 # MODEL_MAX_Length=1024 # if conduct multi-images sft, please set MODEL_MAX_Length=4096

DISTRIBUTED_ARGS=" --nproc_per_node $GPUS_PER_NODE
--nnodes $NNODES
--node_rank $NODE_RANK
--master_addr $MASTER_ADDR
--master_port $MASTER_PORT " torchrun $DISTRIBUTED_ARGS finetune.py
--model_name_or_path $MODEL
--llm_type $LLM_TYPE
--data_path $DATA
--eval_data_path $EVAL_DATA
--remove_unused_columns false
--label_names "labels"
--prediction_loss_only false
--bf16 false
--bf16_full_eval false
--fp16 true
--fp16_full_eval true
--do_train
--do_eval
--tune_vision false
--tune_llm false
--q_lora true
--use_lora true
--lora_target_modules "llm..*layers.\d+.self_attn.(q_proj|k_proj|v_proj|o_proj)"
--model_max_length $MODEL_MAX_Length
--max_slice_nums 9
--max_steps 10000
--eval_steps 1000
--output_dir output/output__lora
--logging_dir output/output_lora
--logging_strategy "steps"
--per_device_train_batch_size 1
--per_device_eval_batch_size 1
--gradient_accumulation_steps 1
--evaluation_strategy "steps"
--save_strategy "steps"
--save_steps 1000
--save_total_limit 10
--learning_rate 1e-6
--weight_decay 0.1
--adam_beta2 0.95
--warmup_ratio 0.01
--lr_scheduler_type "cosine"
--logging_steps 1
--gradient_checkpointing true
--deepspeed ds_config_zero3.json
--report_to "tensorboard" # wandb

以下是我的json文件： train.json conda环境下一些重要的包版本： accelerate 0.30.1
deepspeed 0.14.4 mmengine 0.10.4
mmengine-lite 0.10.4
modelscope 1.17.1
modelscope-studio 0.4.0.9
more-itertools 10.1.0
mpmath 1.3.0
ms-opencompass 0.1.0
ms-swift 2.4.0
ms-vlmeval 0.0.7 opencv-python 4.10.0.84
opencv-python-headless 4.5.5.64 transformers 4.40.0
transformers-stream-generator 0.0.5 torch 2.1.2
torchscale 0.3.0
torchvision 0.16.0

[YifanLiang-hust]

@KeepFaithMe 你好，我在微调时也遇到了这个问题，如果我没记错的话，dataset.py和finetune.py中似乎还有其他地方进行修改（时间久远细节记不太清楚了），才能正常读取数据，建议结合报错信息进行排查

[KeepFaithMe]

非常感谢您的回复 您看是不是将图片中的conversation改为conversations。因为我发现上文根本没有conversation这个变量

[qyc-98]

please try our new finetuning code

[zhangleicv]

有人解决这个问题了吗，在线急求

[beingdutta]

The issue is still there, causing the "Data Fetch Error", in new finetuning code. If anybody fixed it please help.

[whoam-challenge]

怎么解决这个问题呢？求解决方案

[Cuiunbo]

please try our new finetuning code