huggingface/peft issues and pull requests

#1270 - Can peft support singular value decomposition?

Issue - State: closed - Opened by hobbitlzy 10 months ago - 16 comments

#1269 - [docs] Concept guides

Pull Request - State: closed - Opened by stevhliu 10 months ago - 2 comments

#1268 - Refactor and a couple of fixes for adapter layer updates

Pull Request - State: closed - Opened by BenjaminBossan 10 months ago - 4 comments

#1267 - support merge and unload when using PEFT + DeepSpeed ZeRO3

Pull Request - State: closed - Opened by pacman100 10 months ago - 7 comments

#1266 - TST: Revert device_map for AdaLora 4bit GPU test

Pull Request - State: closed - Opened by BenjaminBossan 10 months ago - 1 comment

#1265 - When generate outputs, how to get the probility of the outputs? Is there any param to let the model output probility ?

Issue - State: closed - Opened by ShawnALiu 10 months ago - 1 comment

#1264 - done

Issue - State: closed - Opened by aohan237 10 months ago

#1263 - Fix logic in target module finding

Pull Request - State: closed - Opened by s-k-yx 10 months ago - 14 comments

#1262 - Fix logic in target module finding

Pull Request - State: closed - Opened by s-k-yx 10 months ago

#1261 - FIX Error in log_reports.py

Pull Request - State: closed - Opened by BenjaminBossan 10 months ago - 1 comment

#1260 - This regex is confusing

Issue - State: closed - Opened by s-k-yx 10 months ago - 3 comments

#1259 - PEFT Model returns same results like original model after fine tuning using LORA.

Issue - State: closed - Opened by SachinLimbachiya 10 months ago - 2 comments

#1258 - Bump version to 0.7.2.dev0 post release

Pull Request - State: closed - Opened by BenjaminBossan 10 months ago - 2 comments

#1257 - Release: 0.7.1

Pull Request - State: closed - Opened by BenjaminBossan 10 months ago - 1 comment

#1256 - LoftQ: Allow quantizing models loaded on the CPU for LoftQ initialization

Pull Request - State: closed - Opened by hiyouga 10 months ago - 5 comments

#1255 - FIX: Fix adaption prompt bug with transformers 4.36.0

Pull Request - State: closed - Opened by younesbelkada 10 months ago - 2 comments

#1253 - Retraining Fine tune model

Issue - State: closed - Opened by AliHaider20 10 months ago - 12 comments

#1252 - Issue with transformers 4.36

Pull Request - State: closed - Opened by BenjaminBossan 10 months ago - 17 comments

#1251 - FIX: Truncate slack message to not exceed 3000 chars

Pull Request - State: closed - Opened by BenjaminBossan 10 months ago - 1 comment

#1250 - Revert "FIX Pin bitsandbytes to <0.41.3 temporarily (#1234)"

Pull Request - State: closed - Opened by Titus-von-Koeller 10 months ago - 1 comment

#1249 - About the dtype of trainable params

Issue - State: closed - Opened by hiyouga 10 months ago - 19 comments

#1248 - Type Error: '<=' not supported between instances of 'LoraConfig' and 'int'

Issue - State: closed - Opened by qy1026 10 months ago - 1 comment

#1247 - How to save parameters in prompt_encoder layers in p-tuning?

Issue - State: closed - Opened by lyt719 10 months ago - 1 comment

#1245 - [Question] What is the main difference between "AutoModelForCasualLM" and "PeftModelForCausalLM"?

Issue - State: closed - Opened by daehuikim 10 months ago - 1 comment

#1244 - Added the option to use the corrected scaling factor for LoRA, based on new research.

Pull Request - State: closed - Opened by Damjan-Kalajdzievski 10 months ago - 4 comments

#1243 - Fix: Multiple adapters with bnb layers

Pull Request - State: closed - Opened by BenjaminBossan 10 months ago - 9 comments

#1242 - Bnb integration test tweaks

Pull Request - State: closed - Opened by Titus-von-Koeller 10 months ago - 3 comments

#1241 - TST: Add tolerance for regression tests

Pull Request - State: closed - Opened by BenjaminBossan 10 months ago - 1 comment

#1240 - [LoftQConfig + LoraConfig] throws size matmul mismatch error

Issue - State: closed - Opened by SoundProvider 10 months ago - 8 comments

#1239 - error while loading lora

Issue - State: closed - Opened by ChrisXULC 10 months ago - 2 comments

#1238 - Fix ModulesToSaveWrapper getattr

Pull Request - State: closed - Opened by zhangsheng377 10 months ago - 3 comments

#1236 - [docs] Quantization

Pull Request - State: closed - Opened by stevhliu 10 months ago - 2 comments

#1235 - AttributeError: 'Linear4bit' object has no attribute 'compute_dtype'

Issue - State: closed - Opened by Margaretmmh 10 months ago - 9 comments

#1234 - FIX: Pin bitsandbytes to <0.41.3 temporarily

Pull Request - State: closed - Opened by BenjaminBossan 10 months ago - 1 comment

#1233 - TST: Run regression test in nightly test runner

Pull Request - State: closed - Opened by BenjaminBossan 10 months ago - 1 comment

#1232 - Copy default value for rank_pattern from LoraConfig

Pull Request - State: closed - Opened by qmeeus 10 months ago - 4 comments

#1231 - Copy default value for rank_pattern from LoraConfig

Pull Request - State: closed - Opened by qmeeus 10 months ago

#1230 - Lazy import of bitsandbytes

Pull Request - State: closed - Opened by BenjaminBossan 10 months ago - 2 comments

#1229 - Is there a TaskType just for origin model, not other fixed task.

Issue - State: closed - Opened by EeyoreLee 10 months ago - 2 comments

#1228 - Using Trainer to store checkpoint raise NotADirectoryError.

Issue - State: closed - Opened by ZionDoki 10 months ago - 2 comments
Labels: solved

#1227 - Bump version to 0.7.1.dev0 post release

Pull Request - State: closed - Opened by BenjaminBossan 10 months ago - 1 comment

#1226 - Enable Adapter is not working exactly the same as when the adapter is merged to the Codellama Model

Issue - State: closed - Opened by akhilamartia 10 months ago - 11 comments

#1225 - ModulesToSaveWrapper has no 'weight' attritude

Issue - State: closed - Opened by zhangsheng377 10 months ago - 5 comments

#1224 - [docs] PEFT integrations

Pull Request - State: closed - Opened by stevhliu 10 months ago - 2 comments

#1223 - Whisper V3 PEFT finetuning is broken

Issue - State: closed - Opened by nshmyrev 10 months ago - 6 comments

#1222 - get_peft_model for quantized EsmForSequenceClassification fails due to "only Tensors of floating point dtype can require gradients"

Issue - State: closed - Opened by brianloyal 10 months ago - 3 comments

#1221 - [docs] OFT

Pull Request - State: closed - Opened by stevhliu 10 months ago - 1 comment

#1220 - Quantization docs

Pull Request - State: closed - Opened by dame-cell 10 months ago - 3 comments

#1218 - DOC: Initialization options for LoRA

Pull Request - State: closed - Opened by BenjaminBossan 10 months ago - 1 comment

#1217 - feat: add apple silicon GPU acceleration

Pull Request - State: closed - Opened by NripeshN 10 months ago - 17 comments

#1216 - quantization docs

Pull Request - State: closed - Opened by dame-cell 10 months ago - 1 comment

#1215 - Removed delete_doc_comment.yml workflow

Pull Request - State: closed - Opened by MKhalusova 10 months ago - 3 comments

#1214 - Release: 0.7.0

Pull Request - State: closed - Opened by BenjaminBossan 10 months ago - 1 comment

#1213 - MNT: Delete the delete doc workflows

Pull Request - State: closed - Opened by BenjaminBossan 10 months ago - 1 comment

#1212 - Fix prompt tuning with left padding

Pull Request - State: closed - Opened by SumanthRH 10 months ago - 10 comments

#1211 - [docs] PeftConfig and PeftModel

Pull Request - State: closed - Opened by stevhliu 10 months ago - 1 comment

#1210 - Fix an issue with layer merging for LoHa and OFT

Pull Request - State: closed - Opened by lukaskuhn-lku 10 months ago - 4 comments

#1209 - 'Linear4bit' object has no attribute 'compute_dtype'

Issue - State: closed - Opened by d1pankarmedhi 10 months ago - 17 comments

#1208 - TST: Add tests for 4bit LoftQ

Pull Request - State: closed - Opened by BenjaminBossan 10 months ago - 2 comments

#1205 - Fix model argument issue (#1198)

Pull Request - State: closed - Opened by ngocbh 10 months ago - 4 comments

#1204 - ENH: Enable OFT adapter for mixed adapter models

Pull Request - State: closed - Opened by BenjaminBossan 10 months ago - 4 comments

#1203 - Got error when trying to merge qlora to model

Issue - State: closed - Opened by khiempm209 10 months ago - 8 comments

#1202 - Fix parallel linear lora

Pull Request - State: closed - Opened by zhangsheng377 10 months ago - 1 comment

#1201 - DOC: Update & improve docstrings and type annotations for common methods and classes

Pull Request - State: open - Opened by BenjaminBossan 10 months ago

#1200 - Don't set config attribute on custom models

Pull Request - State: closed - Opened by BenjaminBossan 10 months ago - 1 comment

#1199 - [`torch.compile()`] fails on PEFT LoRA in `diffusers`

Issue - State: closed - Opened by sayakpaul 10 months ago - 15 comments

#1198 - Potentially a bug

Issue - State: closed - Opened by ngocbh 10 months ago - 3 comments

#1197 - Error while working with custom model in trainer

Issue - State: closed - Opened by funger1990 10 months ago - 7 comments

#1196 - [docs] API docs

Pull Request - State: open - Opened by stevhliu 10 months ago - 2 comments

#1195 - DOC: How to configure new transformers models

Pull Request - State: closed - Opened by BenjaminBossan 10 months ago - 2 comments

#1194 - Figure out failing stable diffusion tests

Issue - State: closed - Opened by BenjaminBossan 10 months ago - 2 comments
Labels: bug, help wanted, PRs welcome to address this

#1193 - Support Conv3D layers in LoRA

Pull Request - State: closed - Opened by pacman100 10 months ago - 4 comments

#1192 - model_kwargs["input_ids"].shape[0], peft_config.num_virtual_tokens KeyError: 'input_ids'

Issue - State: closed - Opened by mengqianshuang 10 months ago - 3 comments

#1191 - [docs] Update index and quicktour

Pull Request - State: open - Opened by stevhliu 10 months ago - 1 comment

#1190 - Extend merge_and_unload to offloaded models

Pull Request - State: closed - Opened by blbadger 10 months ago - 10 comments

#1189 - ENH: Different initialization methods for LoRA

Pull Request - State: closed - Opened by BenjaminBossan 10 months ago - 8 comments

#1188 - prefix tuning for encoder-decoder models in peft different from original implementation?

Issue - State: closed - Opened by mickeymickeymonkey 10 months ago - 2 comments

#1187 - Updating multi LoRA inference examples notebook to be usable

Pull Request - State: closed - Opened by lucapericlp 10 months ago - 2 comments

#1186 - Added lora support for phi

Pull Request - State: closed - Opened by umarbutler 10 months ago - 3 comments

#1185 - [`Tests`] Migrate to AWS runners

Pull Request - State: closed - Opened by younesbelkada 10 months ago - 2 comments

#1184 - LoRA applied to encoder and decoder block of a seq-to-seq model

Issue - State: closed - Opened by BSharmi 10 months ago - 3 comments

#1183 - FIX: Wrong use of base layer

Pull Request - State: closed - Opened by BenjaminBossan 10 months ago - 1 comment

#1182 - Toggle inference_mode during training

Issue - State: closed - Opened by LuciferianInk 10 months ago - 2 comments

#1181 - Fix code example in quicktour.md

Pull Request - State: closed - Opened by merveenoyan 10 months ago - 5 comments

#1180 - Using QLoRA to fine-tune BaiChuan-13B reports AttributeError: 'Parameter' object has no attribute 'weight'

Issue - State: closed - Opened by Aitejiu 10 months ago - 2 comments

#1178 - Weird Result between different implementation of LoRA (theoretically should be the same, but the results are drastically different)

Issue - State: closed - Opened by nbasyl 10 months ago - 4 comments

#1175 - Inquiry about the initialization method of adaption_prompt

Issue - State: closed - Opened by callanwu 10 months ago - 3 comments

#1173 - add token for phi-1.5 tokenizer

Issue - State: closed - Opened by lastrei 10 months ago - 5 comments

#1172 - change dtype of output after passing through lora_A

Issue - State: closed - Opened by huseyinatahaninan 10 months ago - 8 comments

#1171 - [docs] Migrate doc files to Markdown

Pull Request - State: closed - Opened by stevhliu 10 months ago - 1 comment

#1170 - Update release checklist about release notes

Pull Request - State: open - Opened by BenjaminBossan 10 months ago - 1 comment

#1169 - fix `add_weighted_adapter` method

Pull Request - State: closed - Opened by pacman100 10 months ago - 2 comments

#1168 - TypeError: PeftConfig.init() got an unexpected keyword argument '_from_model_config'

Issue - State: closed - Opened by GrayZ77 11 months ago - 9 comments

#1167 - Add test for merge_and_unload of offloaded models

Pull Request - State: closed - Opened by blbadger 11 months ago - 7 comments

#1166 - (minor) correct type annotation

Pull Request - State: open - Opened by vwxyzjn 11 months ago - 2 comments

#1165 - Remove explicit declaration of 'label' arg to add support for Causal Models without LM head

Pull Request - State: closed - Opened by huzama 11 months ago - 6 comments

#1164 - Dataset was loaded twice in 4-bit finetuning script

Pull Request - State: closed - Opened by lukaskuhn-lku 11 months ago - 1 comment

#1163 - Mixed adapter models

Pull Request - State: closed - Opened by BenjaminBossan 11 months ago - 2 comments

#1161 - I want to use PEFT to load the Lora model and keep the original model

Issue - State: closed - Opened by LDLINGLINGLING 11 months ago - 2 comments

#1160 - [Feature] Support OFT

Pull Request - State: closed - Opened by okotaku 11 months ago - 16 comments

GitHub / huggingface/peft issues and pull requests