Dedicated Endpoints
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Get help setting up a custom Dedicated Endpoints.
Talk with our engineer to get a quote for reserved GPU instances with discounts.
README
Training
- Checkpoint:
models/sft_temporal_full_h100 - Training data:
data/processed/temporal/train_nisqa_temporal_mix_max_mos3.json - Training job:
sft_temporal_full_h100 - Hardware: 1x H100
- Epochs: 2
- Final train loss: 0.47091
Temporal Test Evaluation
Greedy inference on the temporal test sets:
| Test set | Samples | Parse rate | Mean t-IoU | Median t-IoU | Hit@0.1 | Hit@0.3 | Hit@0.5 |
|---|---|---|---|---|---|---|---|
| FOR | 240 | 1.0000 | 0.2147 | 0.0575 | 0.4792 | 0.3708 | 0.1625 |
| LIVE | 200 | 1.0000 | 0.1261 | 0.0000 | 0.2750 | 0.1850 | 0.1150 |
| P501 | 240 | 1.0000 | 0.2220 | 0.0000 | 0.4458 | 0.3333 | 0.1792 |
Evaluation outputs were produced under:
results/evaluation/temporal/sft_temporal_full_h100_tests.
Model provider
Leng2beat
Model tree
Base
Qwen/Qwen2-Audio-7B
Fine-tuned
this model
Modalities
Input
Audio, Text
Output
Text
Pricing
Dedicated Endpoints
View detailsSupported Functionality
Model APIs
Dedicated Endpoints
Container
More information