Anuar Sharafudinov

OOM with small batch size on GPU

1

Hi, I'm on @lordet01 implementation - https://github.com/lordet01/segan In model.py the following part of code leads to OOM even with small batch_size. Could anyone solve this problem? ``` while not coord.should_stop():...

the best training experience

1

Hello, I have a clean trainset .wav files with following characteristics: bitrate: 128 kb/s, Stream #0:0: Audio: pcm_s16le ([1][0][0][0] / 0x0001), 16000 Hz, 1 channels, s16, 128 kb/s and noisy...

Hi, I am getting GPU out of memory after several epochs of training. It seems like GPU memory is not getting flushed automatically What may be the reason? Thanks in...

Output includes input

5

Here's the code that I'm running ``` tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf", cache_dir=cache_dir) model = AutoModelForCausalLM.from_pretrained("meta-llama/Llama-2-7b-chat-hf", cache_dir=cache_dir) system_message = "You are a helpful, respectful and honest assistant. Always answer as helpfully as...

Добрый день, все работало без перебоев пока не были внесены изменения в default_callback Программа падает регулярно выдавaя разного рода ошибки как "future not running" ``` def default_callback(utterance, start_time = 0.0,...

Hello there, I'm trying to train on 8000 Hz sample rate audio dataset. Is it enough to simply add task.sample_rate=8000 to the fairseq command or there are additional config changes...

OOM with 20GB GPU on SeamlessStreaming evaluate

1

Hi, I needed to transcribe(ASR) 3-9minutes long audios and followed the instructions on https://github.com/facebookresearch/seamless_communication/tree/main/src/seamless_communication/cli/streaming Particularly, by running the following command streaming_evaluate --task asr --data-file test.tsv --audio-root-dir --output --tgt-lang eng --dtype...