-
I'm wondering if it's possible to jailbreak newer models. I tried Llama-3-8B-Instruct without success. Could it be possible with some tweaks to the algorithm or are there some fundamental challenges t…
-
I tried to run the RL training scripts for multiple tasks such as Stabilize, Reach and Grasp, and Insert by
`python3 main/rl/train.py task= sim_device=cuda: rl_device=cuda: graphics_device_id=`
Howe…
-
I got the following error: `the array at index 0 has size 4 and the array at index 1 has size 5` when running the Step2 script in the supervised learning set. My data has 16 images with 2 graph node c…
-
First of all, thank you for sharing the code, and I think you’ve done an amazing job.
However, when I trained your code from scratch, I encountered an issue where center_shift=nan occurred, causing …
-
Hi,
How can I tune the hyperparameters for the model? I couldn't find any reference in this case. Any guidance is appreciated.
-
Hello, I would like to run Meta-Llama-3.1-70B-Instruct on the MATH TEST set. How should I set the system prompt and decoding hyperparameters? Use fewshot or zeroshot?
-
Hello, I want to do some fine-tuning based on your trained model and my own dataset. Since there is no label, I want to replace the coco20k dataset with my own dataset, or add some of my own dataset t…
zw-92 updated
1 month ago
-
I have used random undersample for ratios around 1, but the imbalance situation has great fluctuation in different experiments. Even when the same ratio applies, the number of frames still has differe…
-
This will be to test if the google function returns the correct data completely. Next steps for this would be
Chris
- [ ] add various input generalists to repo
- [ ] add executed responses for …
-
It is currently unclear how best to perform the superresolution aspect of the Popari algorithm. Some pitfalls include:
- Some hyperparameters are not fully exposed on the Popari model object
- It …