Swahili @wheelz86's fine-tuned Whisper large checkpoint managed to get the WER down from 39.3 (large) -> 30.7 Model: https://
huggingface.co/hedronstone/wh
isper-large-v2-sw
… Space:
MACHINE LEARNING
-
Whisper Large Fine-tuned Reduces Swahili Speech Recognition Error
By
–
-
Fine-tuned Whisper Model Achieves Impressive Greek Speech Recognition
By
–
Now for some high-resource languages,
Greek @4re3d0m
's fine-tuned Whisper large checkpoint gets an impressive 10.14 WER when compared to 16 (large) Model: https://
huggingface.co/emilios/whispe
r-large-v2-el-c2
… -
Fine-tuned Whisper Model Achieves 17.8 WER for Latvian
By
–
Latvian @anuragcomm
's fine-tuned Whisper large checkpoint gets an impressive WER of 17.8 when compared to 25.5 by the large checkpoint! Model: -
Whisper Small Model Achieves 8.56 WER for Catalan Speech
By
–
Catalan @jordimash
's fine-tuned Whisper small checkpoint produced an astounding WER of 8.56 compared to 23.8 by the large checkpoint! Model: https://
huggingface.co/softcatala/whi
sper-small-ca
… Space: https://
huggingface.co/spaces/softcat
ala/whisper-demo-catalan
… -
Fine-tuned Whisper Model Achieves Significant WER Improvement for Kannada
By
–
Kannada @saiteya10294's fine-tuned Whisper small checkpoint managed to get the WER down from 100.4 (large) -> 23.1! Model: https://
huggingface.co/steja/whisper-
small-kannada
… -
Lithuanian Whisper Model Achieves Major WER Improvement
By
–
Lithuanian @DeividasMat
's fine-tuned Whisper medium checkpoint managed to get the WER down from 35.2 (large) -> 20.44 Model: -
Whisper Fine-tuning Event Results with LambdaAPI
By
–
10 days into the Whisper fine-tuning event powered by @LambdaAPI and we are well on our descent into the WER land! Here are a select few examples of fine-tuned models and how they compare with zero-shot performance Whisper!
-
Researchers Win OGB-LSC Competition with Accelerated GNN Training
By
–
Super proud of our researchers for winning the MAG240M Track of the OGB-LSC competition at #NeurIPS2022! They used PGL and PaddlePaddle to accelerate GNN training on GPUs, reaching SOTA in only 1 hr (40 hrs before). Check out their code: https://
github.com/PaddlePaddle/P
GL/tree/main/examples/NeurIPS2022-OGB-Challenge/MAG240M
… -
ERNIE-Code: Unified Multilingual Code-to-Text Model
By
–
Introducing ERNIE-Code, a unified pre-trained language model for 116 natural languages and 6 programming languages. ERNIE-Code outperforms previous models for multilingual code-to-text, text-to-code, code-to-code, and text-to-text generation. Paper: https://
arxiv.org/abs/2212.06742 -
How GPT Obtains Its Ability: Tracing Emergent Abilities
By
–
It’s from here: https://
yaofu.notion.site/How-does-GPT-O
btain-its-Ability-Tracing-Emergent-Abilities-of-Language-Models-to-their-Sources-b9a57ac0fcf74f30a1ab9e3e36fa1dc1
…