Issues
- 5
DoRA uses lots of GPU VRAM due to fp32 upcasting
#1692 opened by rationalism - 0
Adapter Merge for Idefics2
#1744 opened by alielfilali01 - 6
TypeError: PeftConfig.__init__() got an unexpected keyword argument '_name_or_path' 报错怎么解决
#1661 opened by caijx168 - 3
FSDP Dora/QDora Broken
#1737 opened by mallorbc - 9
Add Support for IA3 Adapters in add_weighted_adapter Method, Currently facing issue that 'IA3Model' object has no attribute 'add_weighted_adapter'
#1688 opened by Abdullah-kwl - 10
Using PEFT causes model to not predict EOS
#1672 opened by Km3888 - 4
- 5
- 5
- 2
RuntimeError: CUDA error: invalid device ordinal
#1656 opened by yumath - 6
Delete certain layers from PEFT model.
#1730 opened by whr819987540 - 5
RuntimeError: only Tensors of floating point dtype can require gradients for QLoRA since transformers 4.40
#1720 opened by dipanjanS - 0
Reproducibility when using a model with batch norm
#1732 opened by BenjaminBossan - 4
Backward compatibility on saved config.
#1650 opened by xkszltl - 0
DoRA training in distributed setting
#1731 opened by BenjaminBossan - 3
Trainer.train() giving me Key Error: [random number]
#1729 opened by fishroll23 - 2
Add support for OpenELM LoRA fine-tuning
#1726 opened by RonanKMcGovern - 2
[docs] usability
#1645 opened by stas00 - 1
Initialization for LoRA weights A and B initialized
#1728 opened by sanaullah-06 - 3
- 1
- 2
Unable to Install peft==0.7.0.dev0
#1718 opened by AnonymXXXXX - 4
OOM with Phi-3-mini (3.8B) on 83.5GB RAM due to LoftQ
#1708 opened by adamamer20 - 2
- 1
eval_loss showing Nan but train_loss decreases and goes to NaN after couple of steps while fine tuning gemma model with additional vocab
#1715 opened by sidtandon2014 - 1
GPU Allocation Issue (QLoRa + Llama3-8B-IT)
#1716 opened by DONGRYEOLLEE1 - 3
how to finetune whisper model with 'initial_prompt'
#1712 opened by v-yunbin - 5
model merge_and_unload do not support layer_replication
#1707 opened by CrazyBoyM - 2
OOM error while QLoRA+Deepspeed fine tuning of Llama3-70B model on 4xA100-40GB gpus
#1703 opened by hrushikesh198 - 1
Ignore keys for modules to save
#1709 opened by zankner - 2
- 1
Can peft support ColumnParallelLinear?
#1711 opened by wjn1996 - 6
Integrating ReFT
#1654 opened by raven38 - 0
Support merge_and_unload for IA3 Adapters with 4-bit and 8bit Quantization models
#1704 opened by Abdullah-kwl - 14
Saved weights differ from the original model
#1689 opened by bezir - 4
Error while loading PEFT lora model
#1700 opened by Zuhashaik - 2
The FSDP example fails to run with "ValueError: Must flatten tensors with uniform requires_grad when use_orig_params=False"
#1702 opened by xiangxu-google - 1
Outdated utility function: No attribute get_module_class_from_name in FullyShardedDataParallelPlugin
#1699 opened by Xirid - 2
Clarification needed on Adapter Heads in PEFT
#1676 opened by lenglaender - 5
- 1
- 0
Example for Prompt-based methods fails with `expected sequence of length` mismatch
#1685 opened by jpodivin - 1
Problem with merging AdaLoRA adapters
#1647 opened by h4iku - 4
- 5
- 1
- 1
Updating only one adapter using multi adapter
#1670 opened by LameloBally - 1
DoRA support for Embedding
#1677 opened by ShayekhBinIslam - 4
Training with out Trainer class. (VRAM usage issue)
#1651 opened by venzino-han - 2