Ecosyste.ms: Issues

An open API service for providing issue and pull request metadata for open source projects.

GitHub / ssbuild/chatglm_finetuning issues and pull requests

#284 - ptv2

Pull Request - State: closed - Opened by ssbuild about 1 year ago

#283 - num_layers_freeze

Pull Request - State: closed - Opened by ssbuild about 1 year ago

#282 - 简化

Pull Request - State: closed - Opened by ssbuild about 1 year ago

#281 - "gradient_checkpointing": False

Pull Request - State: closed - Opened by ssbuild about 1 year ago

#280 - support accelerator trainer

Pull Request - State: closed - Opened by ssbuild about 1 year ago

#279 - support accelerator trainer

Pull Request - State: closed - Opened by ssbuild about 1 year ago

#278 - v0.2.5

Pull Request - State: closed - Opened by ssbuild about 1 year ago

#277 - v0.2.5

Pull Request - State: closed - Opened by ssbuild about 1 year ago

#276 - support ia3

Pull Request - State: closed - Opened by ssbuild about 1 year ago

#275 - 0.2.4

Pull Request - State: closed - Opened by ssbuild about 1 year ago

#274 - fix slidding

Pull Request - State: closed - Opened by ssbuild about 1 year ago

#273 - update

Pull Request - State: closed - Opened by ssbuild about 1 year ago

#272 - update

Pull Request - State: closed - Opened by ssbuild about 1 year ago

#271 - deepspeed precision

Pull Request - State: closed - Opened by ssbuild about 1 year ago

#270 - fix ptv2

Pull Request - State: closed - Opened by ssbuild about 1 year ago

#269 - fix ptv2

Pull Request - State: closed - Opened by ssbuild about 1 year ago

#268 - ptv2 remove device_map

Pull Request - State: closed - Opened by ssbuild about 1 year ago

#267 - build_template

Pull Request - State: closed - Opened by ssbuild about 1 year ago

#266 - 请问adalora能用deepspeed训练吗

Issue - State: open - Opened by Yu-Yuqing about 1 year ago

#265 - update

Pull Request - State: closed - Opened by ssbuild about 1 year ago

#264 - LoRA和ptv2微调均发生OOM

Issue - State: open - Opened by shenzhyzzz about 1 year ago - 4 comments

#263 - 0.2.0

Pull Request - State: closed - Opened by ssbuild about 1 year ago

#262 - 0.1.21

Pull Request - State: closed - Opened by ssbuild about 1 year ago

#260 - 有谁用过Mac Studio微调的

Issue - State: open - Opened by xsailor511 over 1 year ago

#259 - 怎么控制每训练n轮就保存一次模型呢

Issue - State: closed - Opened by tjulh over 1 year ago - 1 comment

#258 - AttributeError: module 'torch.optim' has no attribute 'adam'

Issue - State: open - Opened by evanweiguohua over 1 year ago - 5 comments
Labels: bug

#257 - 推理时怎么指定用哪几张卡

Issue - State: closed - Opened by tjulh over 1 year ago - 2 comments

#256 - 修改max_seq_length好像并没有生效？

Issue - State: closed - Opened by tjulh over 1 year ago - 4 comments

#255 - AttributeError: module 'inspect' has no attribute 'ArgSpec'

Issue - State: closed - Opened by SeekPoint over 1 year ago - 1 comment
Labels: bug

#254 - 显示可训练参数数量问题

Issue - State: open - Opened by xxll88 over 1 year ago

#253 - 缺省Lora训练显存消耗 60G

Issue - State: open - Opened by is over 1 year ago

#251 - fix potential expand vocab_size

Pull Request - State: closed - Opened by ssbuild over 1 year ago

#250 - requirements.txt

Pull Request - State: closed - Opened by ssbuild over 1 year ago

#249 - load float16 weight

Pull Request - State: closed - Opened by ssbuild over 1 year ago

#248 - support resize embs

Pull Request - State: closed - Opened by ssbuild over 1 year ago

#247 - 模型训练只使用到了单个GPU

Issue - State: closed - Opened by GZJAS over 1 year ago - 1 comment

#246 - 0.1.10

Pull Request - State: closed - Opened by ssbuild over 1 year ago

#245 - ptuning v2 如何启动quantization_bit 4

Issue - State: open - Opened by xxll88 over 1 year ago - 1 comment

#244 - v0.1.10

Pull Request - State: closed - Opened by ssbuild over 1 year ago

#243 - 用单轮数据集。 p-tuning微调chatGLM之后出现的问题。

Issue - State: open - Opened by SMR-S over 1 year ago - 1 comment

#241 - should be load_sft_weight?

Issue - State: closed - Opened by HenryYuxuanWang over 1 year ago - 1 comment
Labels: bug

#237 - input_ids格式是否需要<CLS>

Issue - State: open - Opened by Jong-Won over 1 year ago

#236 - 如何使用evaluate.py对测试集进行验证

Issue - State: open - Opened by lawrencelxy over 1 year ago - 4 comments
Labels: new feature, good issue

#234 - 关于需要多少显卡资源

Issue - State: open - Opened by sanwei111 over 1 year ago - 1 comment

#232 - ptv2显存不够？

Issue - State: open - Opened by sanwei111 over 1 year ago - 11 comments

#231 - 单机两卡指令怎么样

Issue - State: open - Opened by sanwei111 over 1 year ago - 2 comments

#230 - 关于数据的instruction，input，output

Issue - State: open - Opened by sanwei111 over 1 year ago - 3 comments

#229 - v2

Pull Request - State: closed - Opened by ssbuild over 1 year ago

#228 - 关于数据格式

Issue - State: open - Opened by sanwei111 over 1 year ago - 6 comments

#227 - V2 merge

Pull Request - State: closed - Opened by ssbuild over 1 year ago

#225 - v2

Pull Request - State: closed - Opened by ssbuild over 1 year ago

#224 - v2

Pull Request - State: closed - Opened by ssbuild over 1 year ago

#223 - 加载lora模型出错～

Issue - State: closed - Opened by zlht812 over 1 year ago

#222 - merge v2

Pull Request - State: closed - Opened by ssbuild over 1 year ago

#221 - 请问如何试用一般新闻语料对ChatGLM进行继续finetuing呢？

Issue - State: open - Opened by yang9112 over 1 year ago - 1 comment

#220 - web/api_lora_demo.py 如何多张卡推理

Issue - State: open - Opened by lxw0109 over 1 year ago

#217 - 请问一下，mac系统装不了deep_training？

Issue - State: closed - Opened by WHJTC over 1 year ago - 1 comment

#216 - Lora推理2分30s正常吗？

Issue - State: closed - Opened by jikhunb over 1 year ago - 2 comments

#215 - Lora训练后推理问题

Issue - State: closed - Opened by jikhunb over 1 year ago - 2 comments

#215 - Lora训练后推理问题

Issue - State: closed - Opened by jikhunb over 1 year ago - 2 comments

#214 - python train.py执行训练报错，求解。

Issue - State: closed - Opened by pan365wang over 1 year ago - 9 comments

#213 - 设置 LoRa微调的 'target_modules' 后，运行报错 "AssertionError"

Issue - State: closed - Opened by ngbruce over 1 year ago - 4 comments
Labels: wontfix

#212 - Deepspeed stage3保存模型权重维度为0

Issue - State: closed - Opened by Jong-Won over 1 year ago - 2 comments

#212 - Deepspeed stage3保存模型权重维度为0

Issue - State: closed - Opened by Jong-Won over 1 year ago - 2 comments

#210 - 大佬好，请问关于scheduler

Issue - State: closed - Opened by IamRoBota over 1 year ago - 4 comments

#209 - deepspeed如何设置可以避免OOM

Issue - State: open - Opened by lianrzh over 1 year ago - 2 comments

#209 - deepspeed如何设置可以避免OOM

Issue - State: open - Opened by lianrzh over 1 year ago - 2 comments

#208 - 大佬好，请问下数据构造中的特殊token

Issue - State: open - Opened by IamRoBota over 1 year ago - 2 comments

#207 - 数据集

Issue - State: open - Opened by renmengjie7 over 1 year ago

#203 - 大佬 ，能讲一下如何合并lora权重到原来的模型中吗？

Issue - State: closed - Opened by cywjava over 1 year ago - 5 comments

#202 - Lora int8微调，推理时出错

Issue - State: closed - Opened by crellian over 1 year ago - 4 comments

#158 - 有没有大佬试验过哪个更好一些？ptv2和lora参数

Issue - State: closed - Opened by cristianohello over 1 year ago - 1 comment

#146 - 求助 lora load_in_8bit 参数设置

Issue - State: closed - Opened by Zarc98 over 1 year ago - 34 comments
Labels: bug, good issue

#141 - 预计什么时候lora能够支持用deepspeed方式训练

Issue - State: closed - Opened by penguindadyy over 1 year ago - 2 comments
Labels: new feature

#115 - LoRA做infer的时候用int4之后，模型性能会大幅度下降

Issue - State: open - Opened by JamesQFreeman over 1 year ago - 1 comment

#115 - LoRA做infer的时候用int4之后，模型性能会大幅度下降

Issue - State: open - Opened by JamesQFreeman over 1 year ago - 1 comment

#80 - loss不收敛的问题

Issue - State: closed - Opened by weizhenzhao over 1 year ago - 31 comments