Comments (6)
natolambert
commented on June 24, 2024
1
Residual value prediction didn't help with stability (it's crimson-wish)
from trl.
natolambert
commented on June 24, 2024
1
Also not a big help via the other approx KL formulation. W&B here. Though, it's slightly more stable?
We'll see how this run finishes converging.
from trl.
natolambert
commented on June 24, 2024
It could be good to make things like this configurable in a branch and learning how these implementation details transfer to RLHF.
from trl.
DaehanKim
commented on June 24, 2024
imo, residual clipping seems beneficial to prevent policy loss spiking reported in #101 . It's probably coming from instability in value estimation.
from trl.
natolambert
commented on June 24, 2024
Yeah, I'm running residual clipping example(s), we'll see. At least it'll be good to have the option to try both.
from trl.
lvwerra
commented on June 24, 2024
Closing this for now, feel free to reopen if there's an update.
from trl.
Related Issues (20)
- Cannot train custom `PeftModelForCausalLM` model with `PPOTrainer` HOT 1
- KTOTrainer fails to compute loss when model is loaded across multiple GPUs HOT 4
- Support for training LLM's using RLAIF methods HOT 2
- SFT - Dataset Packing causes C10d timeout
- TrlParser not respecting config
- Reproducing LLaVA with Mistral backened
- Bug in calling model.eval() in PPO
- vsft_llava.py:
- Fail to train with ORPOTrainer under multi-GPUs setting HOT 8
- NameError: name 'PeftConfig' is not defined HOT 4
- Loss value returned by the SFT Trainer
- Using PEFT causes model to not predict EOS
- Using SFTTrainer and the DeepSpeed Zero-3 Method training, the data map function runs, but the map function does not run in parallel GPU form. HOT 1
- IPO loss computation vs changes in DPO dataset format
- Anthropic HH new dataset format repeats the prompt HOT 1
- Group Relative Policy Optimization Trainer
- bugs in trained save and evaluation for DPO training with deepspeed_zero3 HOT 3
- config.json file is missing. HOT 3
- The data is consuming too much GPU memory In the SFTTrainer HOT 2
- [Question] Does TRL support DPO trainer with simulated environment for generating training data like PPO's step-based training?
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 trl.