johnsmith0031 alpaca_lora_4bit issues

johnsmith0031 / alpaca_lora_4bit

MIT License

533 stars 86 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

Support for moe model？

#156 laoda513 opened 3 months ago
2
Why lora support is only for simple lora with only q_proj and v_proj ？

#155 XpracticeYSKM opened 4 months ago
1
trying to get this working with text-generation-webui

#154 caterpillarpants opened 8 months ago
9
AttributeError: 'dict' object has no attribute 'to_dict'

#153 kkaarrss closed 10 months ago
1
Finetuning CodeLLaMA34B - RuntimeError: The size of tensor a (1024) must match the size of tensor b (8192)

#152 juanps90 closed 10 months ago
3
docker.io/nvidia/cuda:11.7.0-devel-ubuntu22.04 not available anymore

#151 kkaarrss closed 10 months ago
1
3 errors detected in the compilation of "src/alpaca_lora_4bit/quant_cuda/quant_cuda_kernel.cu"

#150 kkaarrss closed 10 months ago
2
Refactor generate_prompt to f-strings and make input optional

#149 jordankzf closed 10 months ago
1
ValueError: Target module Autograd4bitQuantLinear() is not supported.

#148 jordankzf closed 10 months ago
7
module 'alpaca_lora_4bit.quant_cuda' has no attribute 'vecquant4recons_v2'

#147 kevkid closed 10 months ago
4
OOM on inference while i can finetune with more tokens

#146 nepeee closed 10 months ago
2
Merging LoRA after finetune

#145 gameveloster opened 11 months ago
1
Unable to Build Wheels

#144 VegaStarlake closed 11 months ago
8
Is alpaca_lora_4bit@winglian-setup_pip missing finetune.py?

#143 tensiondriven opened 12 months ago
1
Feature request: Stop when loss reaches X

#142 tensiondriven opened 12 months ago
1
Targeting all layers and biases

#141 grimulkan closed 11 months ago
2
High perplexity while lower loss after LoRA finetuning (how?)

#140 alex4321 closed 12 months ago
5
Error attempting to finetune llama2-70b

#139 tensiondriven opened 12 months ago
5
Flash Attention 2

#138 ghost closed 12 months ago
1
LoRA Output Identical to Base Model

#137 LegendBegins closed 12 months ago
4
How to use inference.py after finetune.py?

#136 athenawisdoms closed 1 year ago
2
Checkpoint saving broken with the latest version of huggingface

#135 nepeee closed 11 months ago
8
Tests

#134 alex4321 closed 1 year ago
1
Use PyPI versions of dependencies

#133 alex4321 closed 1 year ago
1
July

#132 ghost closed 1 year ago
4
Crashes during finetuning

#131 gameveloster opened 1 year ago
2
TypeError: object of type 'NoneType' has no len()

#130 athenawisdoms closed 1 year ago
1
Update docs for > 2048 token models (SuperHOT)?

#129 tensiondriven opened 1 year ago
12
Use gradient checkpoint only for training mode, not evaluation

#128 alex4321 closed 1 year ago
2
Gibberish results for non-disabled "faster_mode" using "vicuna-7B-GPTQ-4bit-128g" model

#127 alex4321 opened 1 year ago
4
Zero initializer for biases

#126 alex4321 closed 1 year ago
1
Fix NaN or Inf after initializing Vicuna models (due to lack of bias weights)

#125 alex4321 closed 1 year ago
1
Inf or NaN in probabilities. Windows 10, vicuna-7b-gptq-4bit-128g

#124 alex4321 closed 1 year ago
35
Remove import of custom_autotune (it's triton-dependent, triton depen…

#123 alex4321 closed 1 year ago
2
this repo support 2bit finetuning the llama model？ Is there any case to show how to run the scripts?

#122 zlh1992 opened 1 year ago
1
[question] weights in the replaced quantized modules

#121 vince62s closed 1 year ago
0
how to change into 8 bit

#120 leexinyu1204 opened 1 year ago
1
Problem with inference

#119 leexinyu1204 closed 1 year ago
7
fine tune with 2 GPU

#118 shawei3000 opened 1 year ago
2
how to infer with finetuned model?

#117 balaji-skoruz opened 1 year ago
4
Implementing Landmark Attention

#116 juanps90 opened 1 year ago
0
Finetuning 2-bit Quantized Models

#115 kuleshov opened 1 year ago
7
ImportError: cannot import name '_get_submodules' from 'peft.utils'

#114 saber258 closed 1 year ago
10
Differences between QLoRA and this repo

#113 qwopqwop200 opened 1 year ago
3
Code reference request

#112 PanQiWei opened 1 year ago
1
what is the difference between v1 model and v2 model?

#111 zlh1992 opened 1 year ago
1
Problem loading safetensor file format

#110 ortegaalfredo opened 1 year ago
3
Consider using new QLoRA

#107 juanps90 opened 1 year ago
3
ValueError: Autograd4bitQuantLinear() does not have a parameter or a buffer named qzeros.

#105 ra-MANUJ-an opened 1 year ago
2
Version of GPTQ

#104 juanps90 opened 1 year ago
3