-
Hello,
Beam later version is V2 and they did drastic changes to their SDK and client that makes most of the training (fine-tuning) and inference code useless. There is no "beam run" and so on...
…
-
## 🐛 Bug
Not sure if this is a feature request or bug. I took the [SPMD Gemma ft code from Hugging Face](https://huggingface.co/google/gemma-7b/blob/main/examples/example_fsdp.py) and tried to run …
-
### Have I written custom code (as opposed to using a stock example script provided in MediaPipe)
None
### OS Platform and Distribution
IOS
### MediaPipe Tasks SDK version
_No response_
### Task…
-
I tried to use ctranslate2 as the inference framework to do model inference, but failed with error as below:
"axis 2 has dimension 8192 but expected 7680"
What I've done:
1. First I must con…
-
# Prerequisites
Please answer the following questions for yourself before submitting an issue.
- [ Yes] I am running the latest code. Development is very rapid so there are no tagged versions as…
-
### Describe the bug
1) your ui initial start is like this (perfect):
2) whenever you open additional inputs and then closed then height end like this:
in terms of UX is no functional.
…
-
I tried to reproduce your gemma2B reward model training again and found that the reward model architecture fine tuned with internlm2 had an output header of 1. I downloaded your GRM-Gemma-2B-Sftrug re…
-
Hello Authors,
Thank you for your incredible work and the comprehensive experiments presented in the paper.
I have a question regarding the implementation of attacks. Specifically, some attacks,…
-
Since the latest models, such as Llama 3 and Gemma, adopt extremely large vocabularies (128-256K), the size of logits can become very large, consuming a large proportion of VRAM. For example, the foll…
-
Compile the database of learning materials that contains longer and more reasonable materials.