Downloads · 30 days
29
23% of all-time downloads
KasuleTrevor/cdli-qwen3-asr-lg-atypical-stage3-1p7b-typical-base-const5e5-fresh
cdli-qwen3-asr-lg-atypical-stage3-1p7b-typical-base-const5e5-fresh is a automatic speech recognition model from KasuleTrevor. Use it when you need speech turned into text. It is set up for transformers. The card lists the license as apache-2.0.
This repo contains the selected checkpoint checkpoint-3500 from the LG-QWEN3-ASR-ATYPICAL-STAGE3-1P7B-TYPICAL-BASE-CONST5E5-FRESH run.
Downloads · 30 days
29
23% of all-time downloads
All-time downloads
128
Public
Parameters
2B
4.1 GB on disk
Likes
0
Public
Click a slice to open those files.
.safetensors4.1 GB · 100%
From the Hugging Face model README
This repo contains the selected checkpoint checkpoint-3500 from the LG-QWEN3-ASR-ATYPICAL-STAGE3-1P7B-TYPICAL-BASE-CONST5E5-FRESH run.
train_cleaned.tsv, validation_cleaned.tsv, test_cleaned.tsvLuganda| checkpoint | wer_normalized | cer_normalized | avg_wer_normalized | avg_cer_normalized | eval_loss |
|---|---|---|---|---|---|
| checkpoint-3500 | 0.5658713254445168 | 0.28529362421476356 | 0.5658713254445168 | 0.28529362421476356 | 0.7408768534660339 |
| checkpoint-3685 | 0.593422535641947 | 0.2993799972366213 | 0.593422535641947 | 0.2993799972366213 | 0.7408768534660339 |
| severity_speech_impairment | n_samples | n_speakers | mean_wer | mean_cer | median_wer | median_cer |
|---|---|---|---|---|---|---|
| Severe (frequent breakdowns) | 315 | 3 | 0.5503 | 0.2755 | 0.5714 | 0.2561 |
| Moderate (requires effort to understand) | 347 | 3 | 0.4314 | 0.1653 | 0.4 | 0.1256 |
| Mild (easily understood with minimal effort) | 365 | 3 | 0.3482 | 0.1246 | 0.3333 | 0.0732 |
results/checkpoint-3500/