EssentialAI/rnj-1-instruct
Text Generation
•
8B
•
Updated
•
15.6k
•
•
293
Terra Nulles
TerraNull
·
AI & ML interests
None yet
Recent Activity
liked
a model
about 1 month ago
EssentialAI/rnj-1-instruct
liked
a model
5 months ago
openai/gpt-oss-120b
liked
a model
10 months ago
Qwen/QwQ-32B
Organizations
None yet
liked
a
model
about 1 month ago
liked
a
model
5 months ago
liked
a
model
10 months ago
upvoted
a
paper
11 months ago
liked
2
models
about 1 year ago
liked
a
model
over 1 year ago
reacted to
ehristoforu's
post with 🤝
over 1 year ago
Post
2149
😐 Hello, there are a couple of interesting things. The first is that I will soon release several pretty cool SDXL models, the second is a little sad, I conducted long-term tests of training and merging of XL models and realized that XL will not improve soon, the architecture will not allow us to continue pushing realism and other interesting things into it, the entire community has brought XL closer to the maximum ideal on its architecture.
liked
a
Space
over 1 year ago
liked
a
model
over 1 year ago
Feature Request for Leaderboard: date added to hub
👍
4
8
#425 opened about 2 years ago
by
madmaxbr5
liked
a
model
over 1 year ago
liked
a
model
almost 2 years ago
liked
a
Space
almost 2 years ago
reacted to
vladbogo's
post with 👍
almost 2 years ago
Post
"A Closer Look at the Limitations of Instruction Tuning" is a new paper that explores the efficacy and limitations of Instruction Tuning (IT) in Large Language Models (LLMs) for conversational agents. The authors conduct a series of experiments using both LoRA fine-tuning (LFT) and standard full-parameter fine-tuning (SFT) across various LLMs and IT datasets.
The key findings are:
* LoRA fine-tuning (LFT) preserves the pre-training token distribution while SFT doesn't. This indicates that using LFT, post fine-tuning the model still heavily relies on the pre-training and doesn't acquire new information.
* Dataset scaling is ineffective for LFT - experiments show that scaling the dataset size 52x or even 326x doesn't improve the performance.
* LoRA fine-tuning mainly enhances response initiation and style without substantial knowledge enhancement.
* Full-parameter fine-tuning tends to degrade LLM knowledge base and increase hallucination occurrences.
* Popular other methods and adjustments fail to significantly outperform simple LoRA fine-tuned models in terms of conversational quality and accuracy.
Congrats to the authors @Sreyan88 and others for their work!
Paper: A Closer Look at the Limitations of Instruction Tuning (2402.05119)
The key findings are:
* LoRA fine-tuning (LFT) preserves the pre-training token distribution while SFT doesn't. This indicates that using LFT, post fine-tuning the model still heavily relies on the pre-training and doesn't acquire new information.
* Dataset scaling is ineffective for LFT - experiments show that scaling the dataset size 52x or even 326x doesn't improve the performance.
* LoRA fine-tuning mainly enhances response initiation and style without substantial knowledge enhancement.
* Full-parameter fine-tuning tends to degrade LLM knowledge base and increase hallucination occurrences.
* Popular other methods and adjustments fail to significantly outperform simple LoRA fine-tuned models in terms of conversational quality and accuracy.
Congrats to the authors @Sreyan88 and others for their work!
Paper: A Closer Look at the Limitations of Instruction Tuning (2402.05119)
- 2 replies
upvoted
2
papers
almost 2 years ago
liked
a
model
almost 2 years ago
liked
a
Space
almost 2 years ago
liked
a
model
almost 2 years ago
