Comments (4)
Hi, thanks for your interest.
We have tried our method on llama2-7b and 13b, it works fine.
Our method is built on a widely accepted hypothesis that LLM has learned all the knowledge it needs and it only needs to learn the alignment from instruction tuning. Thus we believe as long as the other LLMs satisfy this hypothesis, it should work fine.
As for Baichuan, I don't know much about this model, and I don't find a lot of instruction-tuned models based on it, the only one that I found is Baichuan-13B-Chat which has a win rate of 21.80%.
Are there any other models based on it?
from cherry_llm.
Thanks for your reply, now I am using the Baichaun2-13B-Chat model for validation on the "alpaca_data.json" dataset. Or can you tell me about other general methods that can automatically filter the high-quality fine-tuning data required by large models during the process of command fine-tuning on large models.
from cherry_llm.
I am also curious about how Baichaun2-13B-Chat will perform on the current commonly used instruction tuning dataset. If possible, please keep me updated~
I have to say I haven't seen many other methods that can be used to automatically filter the high-quality instruction tuning data due to its special properties. There are methods using chatGPT as the filter, but I don't think that is what you need. Sorry about that.
from cherry_llm.
Thanks for your reply, now I am using the Baichaun2-13B-Chat model for validation on the "alpaca_data.json" dataset. Or can you tell me about other general methods that can automatically filter the high-quality fine-tuning data required by large models during the process of command fine-tuning on large models.
Hello! May I ask if you have successfully implemented high-quality data filtering for the Cherry project using Baichaun2-13B Chat? When I called Baichaun2-7B Chat, an error occurred as shown in the following figure.
from transformers import AutoTokenizer,AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained(args.model_name_or_path, device_map="auto",
trust_remote_code=True, torch_dtype=torch.float16)
tokenizer = AutoTokenizer.from_pretrained(args.model_name_or_path,
trust_remote_code=True, torch_dtype=torch.float16)
model.eval()
from cherry_llm.
Related Issues (20)
- a confusion about Instruction-Following Difficulty (IFD) scores HOT 2
- a confusion about data_by_IFD HOT 3
- Logic behind IFD score HOT 1
- I plan to apply this method on Llama2, which part of this project needs to be changed to adapt to Llama2? HOT 1
- about the paper HOT 1
- Multi-round conversation data set HOT 3
- GPT-4/ChatGPT Evaluation Code HOT 1
- How to filter code SFT data? HOT 2
- Questions related to training HOT 5
- Could the Pre-Experienced Model be used in other different dataset? HOT 1
- Any report of time consuming? HOT 1
- Chinese SFT data cannot be displayed. HOT 3
- 'The training of pre-experienced models is discarded for more efficient usage': that means we can only use base model to do cherry analysis and selection? HOT 1
- batch? HOT 1
- 关于Direct Answer Score sθ(A) HOT 2
- Evaluation reproducibility on benchmarks HOT 4
- how many epochs to train on cherry data? HOT 2
- Question about the effect of labels[0, :start_token] = -100 HOT 1
- why is the process so slow HOT 5
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 cherry_llm.