huggingface/candle
1
0
Fork 0
mirror of https://github.com/huggingface/candle.git synced 2025-06-15 10:26:33 +00:00
Code Issues Packages Projects Releases Wiki Activity

Use the same padding in metavoice as in the python version. (#1794)

Browse Source
This commit is contained in:
Laurent Mazare
2024-03-03 12:04:48 +01:00
committed by GitHub
parent 09e0148cce
commit ca5d727ba2
1 changed files with 6 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
candle-examples/examples/metavoice/main.rs
View File

@ -129,7 +129,7 @@ fn main() -> Result<()> {
VarBuilder::from_mmaped_safetensors(&[second_stage_weights], DType::F32, &device)?
};
let second_stage_config = gpt::Config::cfg1b_v0_1();
let second_stage_model = gpt::Model::new(second_stage_config, second_stage_vb)?;
let second_stage_model = gpt::Model::new(second_stage_config.clone(), second_stage_vb)?;
let encodec_device = if device.is_metal() {
&candle::Device::Cpu
@ -182,13 +182,16 @@ fn main() -> Result<()> {
let mut rng = rand::rngs::StdRng::seed_from_u64(args.seed + 1337);
// TODO: Use the config rather than hardcoding the offset here.
let encoded_text: Vec<_> = prompt_tokens.iter().map(|v| v - 1024).collect();
let hierarchies_in1 = [encoded_text.as_slice(), ids1.as_slice(), &[ENCODEC_NTOKENS]].concat();
let hierarchies_in2 = [
let mut hierarchies_in1 =
[encoded_text.as_slice(), ids1.as_slice(), &[ENCODEC_NTOKENS]].concat();
let mut hierarchies_in2 = [
vec![ENCODEC_NTOKENS; encoded_text.len()].as_slice(),
ids2.as_slice(),
&[ENCODEC_NTOKENS],
]
.concat();
hierarchies_in1.resize(second_stage_config.block_size, ENCODEC_NTOKENS);
hierarchies_in2.resize(second_stage_config.block_size, ENCODEC_NTOKENS);
let in_x1 = Tensor::new(hierarchies_in1, &device)?;
let in_x2 = Tensor::new(hierarchies_in2, &device)?;
let in_x = Tensor::stack(&[in_x1, in_x2], 0)?.unsqueeze(0)?;