Comments (4)
@raywang4 Thanks for your interest on our work! May I know if your problem has been solved?
from lorahub.
@SivilTaram Hi there! Thanks for the reply! I tried with the FLAN 2022 dataset but it still seems like my evaluations are quite off. Would you mind specifying what part of the dataset was used for fine-tuning the lora modules? I read from other issues that a min(10k, data size) samples are selected. Can you let me know which template type was used? Thanks a lot!
from lorahub.
@raywang4 Hello! We do not specify the template types, and just try to sample from these datasets using the following code:
from datasets import load_dataset
import os
dataset_folder = "flan_task"
def download_flan():
dataset = load_dataset("conceptofmind/FLAN_2022", split="train")
# filter some examples from the dataset
dataset = dataset.filter(lambda example: example['template_type'] == "zs_noopt", num_proc=32)
# group the dataset using the task_name
task_names = dataset.unique("task_name")
for task_name in task_names:
print("Processing task: ", task_name)
# filter the dataset for the current task
task_dataset = dataset.filter(lambda example: example['task_name'] == task_name, num_proc=32)
# if the dataset is too large, we randomly sample 10000 examples for the training
if len(task_dataset) > 10000:
task_dataset = task_dataset.shuffle()
task_dataset = task_dataset.select(range(10000))
# save it into the task file
task_name = task_name.replace("/", "_")
task_dataset.to_json(os.path.join(dataset_folder, task_name + ".json"))
if __name__ == "__main__":
download_flan()
from lorahub.
Thanks for the code! It's very helpful!
from lorahub.
Related Issues (20)
- Is there any scalability issue? HOT 2
- About training examples HOT 1
- About LoraHub adapters HOT 3
- Request for Guidance on Reproducing Experiments for BigBenchHard HOT 8
- Bug Report - (batched option & tensor shape) HOT 4
- No single LoRA model can score on Disambiguation HOT 5
- Where is the instruction of each lora in lorahub? HOT 1
- Training Tasks and Data for these LoRA modules? HOT 8
- can lorahub be used in non llm tasks HOT 3
- Release of Code for Finetuning Lora modules of FLAN and evaluating the Finetuned Models. HOT 1
- FLAN datasets not available on Huggingface. HOT 10
- repaired an error in reproduce_bbh.py (shuffle seed) HOT 2
- train_model model load repaired HOT 4
- 404 HOT 2
- AttributeError: 'GenerationConfig' object has no attribute 'cache_implementation' HOT 4
- Downloading the FLAN-v2 dataset HOT 1
- Using with CausalLM models HOT 3
- How to evaluate glue lora?
- How to use local LoRA trained with train_model.py HOT 1
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from lorahub.