hexgrad/Kokoro-82M
Text-to-Speech
โข
Updated
โข
129k
โข
โข
2.76k
Kh
raidhon
AI & ML interests
Fine-tuning, Dataset creation, Time Series
Recent Activity
liked
a model
12 days ago
hexgrad/Kokoro-82M
liked
a model
13 days ago
bytedance-research/UI-TARS-72B-DPO
Organizations
None yet
raidhon's activity
liked
a
model
12 days ago
liked
a
model
13 days ago
reacted to
onekq's
post with ๐ฅ
15 days ago
Post
4665
๐DeepSeek ๐ is the real OpenAI ๐ฏ
- 6 replies
liked
a
dataset
22 days ago
Can't reproduce the evaluation result of GPQA dataset
5
#47 opened about 2 months ago
by
Rinn000
liked
a
model
4 months ago
liked
a
dataset
5 months ago
Yes, it's been tested, and it's false. It's even worse than the regular LLAMA 3.1 70b. It's even funny to compare it to Claude.
https://www.reddit.com/r/LocalLLaMA/s/BH5A2ngyui
liked
2
models
9 months ago
replied to
hrishbhdalal's
post
9 months ago
Yeah, I was thinking the same thing. A large vocabulary does improve the performance of smaller LLMs and judging by the GPT-4o the same is true for larger LLM. Give it a try. I'm just doing this for small size models up to 3B parameters.
liked
a
model
9 months ago
updated
a
model
9 months ago
liked
a
model
9 months ago
updated
a
model
9 months ago
updated
a
collection
9 months ago
updated
a
model
9 months ago
updated
a
collection
9 months ago
updated
a
model
9 months ago
updated
a
collection
9 months ago