Deepanway

@ChasonLee , @huydan , @Muugii-bs Hey guys, I have made some modifications in the code so that further predictions can be made on some test examples. You can found it...

@Muugii-bs I don't actually save the parameters. Along with train and validation set I also pass the test set in the function train_conv_net() and it returns predicted test labels. In...

Adding @nmder to this thread as he is the author of the DialogueRNN code.

Please follow the steps here for installation: https://github.com/AMAAI-Lab/mustango#installation

Hi, Flacuna is a LoRA-based model. You can refer to the [flacuna.py](https://github.com/declare-lab/flacuna/blob/main/flacuna.py) file to see how we load the weights in `self.model`.

Can you add the warning message here? The warning you are encountering should also show the parameter names for which the pre-trained weights were not loaded. Also, assuming you have...

This warning should come when you try to initialize `LlamaForCausalLM`directly from `"declare-lab/flacuna-13b-v1.0"`. In flacuna.py, we first initialize Llama from `"TheBloke/vicuna-13B-1.1-HF"` and then load the LoRA weights from our checkpoint. Otherwise,...

Hey, you can try the following: 1. Use a smaller text encoder and a smaller diffusion model if you are training from scratch. 2. Use the Adafactor / 8 Bit...

Yes, you can use the `--hf_model` argument to pass the tango-full model checkpoint for doing that. The full command would be: ``` accelerate launch train.py \ --train_file="data/train_audiocaps.json" --validation_file="data/valid_audiocaps.json" --test_file="data/test_audiocaps_subset.json" \...

The `inferece_hf.py` will compute the evaluation metrics (reported in the paper) from the checkpoints we uploaded in huggingface. For each text prompt, only one audio sample will be generated. The...