-
In [modify_llama.py](https://github.com/FMInference/H2O/blob/main/h2o_hf/utils_real_drop/modify_llama.py), the hh_score of H2OCache is computed by attn_scores.sum(0).sum(1), resulting in a shape of [n…
-
There are no train_choices and val_choices.json files in the GQA dataset.
'train_choices': self.DATA_PATH['gqa'] + '/raw' + '/eval/train_choices'
'val_choices': self.DATA_PATH['gqa'] + '/raw' + '/ev…
QA-x updated
3 months ago
-
- [VQAv2](https://arxiv.org/pdf/1612.00837v3)
- [TallyQA: Answering Complex Counting Questions](https://arxiv.org/pdf/1810.12440)
- [GQA: A New Dataset for Real-World Visual Reasoning and Compos…
-
Hi, great work!
Can you provide checkpoint for GQA model?
Best,
a
-
Hi! a very nice work!
Do you reproduce the Llava results in your project?
-
Thanks for sharing the repository.
Could you please suggest, how to use class GQASceneGraphsOnlyDataset(data.Dataset[NSMItem]) in https://github.com/gchaperon/neural-state-machine/blob/main/nsm/datas…
-
Thanks a lot for your excellent job. I wonder how you evaluate the trained model, do you use ./scripts/more/eval/pope.sh, which uses llava.eval.model_vqa_loader for evaluation (seems no modification f…
-
Platforms: rocm
This test was disabled because it is failing on main branch ([recent examples](https://torch-ci.com/failure?failureCaptures=%5B%22test_transformers.py%3A%3ATestSDPACudaOnlyCUDA%3A%3…
-
In theory, MQA/GQA can reduce memory bandwidth for reading KV cache and enable using TensorCore for the dot products in attention mechanism. However, this benefit can be only realized when using optim…
-
May I ask if this tool is currently unable to perform pruning on GQA models? Llama2-70B or Llama3