hkust-nlp deita issues - Githubissues

hkust-nlp / deita

Deita: Data-Efficient Instruction Tuning for Alignment [ICLR2024]

Apache License 2.0

419 stars 26 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

What is the significance of the id2score function?

#29 zhengjie-zhou opened 5 days ago
0
Computational cost of the algorithm

#28 inigopm opened 1 month ago
0
Could you please publish the original data pool?

#27 ShadowTinker closed 2 months ago
2
The length of samples

#26 Ber666 opened 2 months ago
1
If I want to use vllm, which version should I install

#25 kissshhot opened 3 months ago
0
Does the EVOL process of instruction dataset has been released?

#24 dsj96 opened 3 months ago
0
reproduce mt-bench score

#23 bpucla opened 4 months ago
1
Have you conducted ablation experiments with three factors: Complexity, Quality, and Diversity? Which one has the greatest impact on performance improvement?

#22 447428054 opened 4 months ago
1
Questions about performance improvement in Open LLM leaderboard

#21 minstar opened 4 months ago
3
Cosine distance computation

#20 sangkeun00 closed 4 months ago
4
Question about which score to ultimately use for the filtering process.

#19 447428054 closed 4 months ago
2
Scorer models on hub are 7b not 13b

#18 edbeeching closed 4 months ago
2
fix num_proc

#17 VPeterV closed 5 months ago
0
Can we support more languages?

#16 zhangfan-algo opened 5 months ago
1
[Question] Regarding the order bias in sample scoring.

#15 liujuncn closed 4 months ago
2
fix redundant results in last-batch data

#14 VPeterV closed 5 months ago
0
Dev to main

#13 VPeterV closed 5 months ago
0
Questions about the "Pool=50K" in your paper.

#12 DLiquor closed 4 months ago
2
[Question] Script to train Scorer model?

#11 agi-piggy closed 4 months ago
1
[question]Did you use the mean value of all token embedding in repr filter?

#10 Force1ess closed 6 months ago
3
[question] repr_filter encoding only the instruction or instruction and anwser?

#9 Force1ess closed 6 months ago
1
What content is encoded when Llama13B encoded a sentence

#8 cgpeter96 closed 6 months ago
3
[Question] Is the 6k dataset is a subset of 10k dataset.

#7 ChenMnZ closed 6 months ago
1
make selection.scorer traversable

#6 winglian closed 6 months ago
0
Some questions about running the scorer for arbitary model

#5 HelloWorldLTY closed 6 months ago
2
data of deita's dpo+sft

#4 jiezhangGt closed 6 months ago
0
How did you train the complexity & quality scorer

#3 philschmid closed 6 months ago
8
request: is code of "Repr Filter" can be open source too?

#2 Force1ess closed 6 months ago
4
Update README.md

#1 eltociear closed 6 months ago
0