Comments (2)
Thanks @pascal-pfeiffer for the explanation, that makes sense. Will try with LORA for now on top of an SFT step and see if that yields improvements.
from h2o-llmstudio.
Thank you for raising the request.
DPO needs to compare the logits of the fine-tune with the base logits. For that we leverage LoRA, as we can easily disable and enable the extra layers without needing to have two copies of the same model in the memory. Allowing full tunes, would require twice as much GPU memory and additional code complexity which we believe is not worth the potential gain.
We suggest to do an SFT fine-tune first and use DPO as a second step. The need for full fine-tuning usually stems from a big difference to the pre-training. Doing SFT first, and then DPO removes this need from DPO.
from h2o-llmstudio.
Related Issues (20)
- [FEATURE]create distributed_inference.sh with deepspeed HOT 3
- [CODE IMPROVEMENT] if user specifies a βsystemβ column but it doesnt exist, it should error out instead of continue running silently HOT 5
- [BUG] Unable to access LLM studio using Docker from a nightly build HOT 2
- [BUG] Default dataset can not be written to the database
- [FEATURE] MoE Aux Loss
- CUDA out of memory HOT 6
- [FEATURE] Add UI tests to .github actions
- [CODE IMPROVEMENT] Create CICD for UI Testing HOT 1
- [DOC] Update the code examples to demonstrate how to implement experiments from CLI using YAML configuration files
- Can you kindly provide documentation for macOS HOT 1
- Inference model on distributed GPU after model training via CLI HOT 3
- [DOCS] Adding the description syntax
- [BUG] Recent regression causing an error when loading tokenizer for Deepseek models HOT 2
- [BUG] Mixed precision not working with bfloat16 HOT 2
- [CODE IMPROVEMENT] Use build-in H2O Wave Chatbot features
- [FEATURE] Training with QLoRA + FSDP
- [CODE IMPROVEMENT] Resolve High Vulnerabilities HOT 1
- [BUG] Error when pushing model to HuggingFace HOT 2
- [FEATURE] Option to add BOS token HOT 4
- [BUG] Downloading a fine-tuned model fails with error HOT 2
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
π 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. πππ
-
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 β€οΈ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from h2o-llmstudio.