Hugging Face's logo

dd3434
/
Streamalign-SLM-R16

Model card Files
xet
 Community
dd3434 commited on
Commit
e23f8ee
·
verified ·
1 Parent(s): 0f91744

Use role-based names instead of internal experiment run names

Browse files
Files changed (1) hide show
  1. README.md +1 -1
README.md CHANGED
@@ -4,6 +4,6 @@ StreamAlign speech language model (Llama-3.2-1B backbone, R16 hier
4
  durfirst-durreg, text-KL recipe). Scores speech via the paired R16 RVQ
5
  teacher `Streamalign-R16`.
6
 
7
- - Original checkpoint: `R16_step180k_lossWtext5_kl05_210k/step_00185000.pt`
8
  - Speech tokenizer / teacher: `Streamalign-R16`
9
  - Reported metrics: SALMon 69.1 (4-task, no pitch), StoryCloze 72.1 (sSC 59.0 / tSC 85.1)
 
4
  durfirst-durreg, text-KL recipe). Scores speech via the paired R16 RVQ
5
  teacher `Streamalign-R16`.
6
 
7
+ - Original checkpoint: `streamalign_slm_r16/step_00185000.pt`
8
  - Speech tokenizer / teacher: `Streamalign-R16`
9
  - Reported metrics: SALMon 69.1 (4-task, no pitch), StoryCloze 72.1 (sSC 59.0 / tSC 85.1)