o3-mini vs Deepseek-R1
•
13
Prithiv Sakthi PRO
prithivMLmods
AI & ML interests
computer vision, multimodality, realism engine adapters @starngerzonehf
Recent Activity
new activity
about 4 hours ago
prithivMLmods/Calcium-Opus-14B-Elite2-R1:Adding Evaluation Results
updated
a model
about 4 hours ago
prithivMLmods/Calcium-Opus-14B-Elite2-R1
updated
a model
about 20 hours ago
prithivMLmods/LatexMind-2B-Codec
Articles
Organizations
prithivMLmods's activity
posted
an
update
about 21 hours ago
Post
1194
o3-Mini and Deepseek R1
Worked out with some famous and weird examples.
🔥Blog: https://huggingface.co./blog/prithivMLmods/o3-mini-vs-deepseek-r1
Prompt : Using HTML, CSS, and JavaScript in a single HTML file to create a simulation of the solar system. Pay extreme attention to the UI to make it as intuitive as possible. Ensure that every planet appears as a sphere and is labeled with its corresponding name.
example 1: o3 Mini , example 2: Deepseek R1
Q2 : https://huggingface.co./blog/prithivMLmods/o3-mini-vs-deepseek-r1#q2--web-solar-system-explorer
Worked out with some famous and weird examples.
🔥Blog: https://huggingface.co./blog/prithivMLmods/o3-mini-vs-deepseek-r1
Prompt : Using HTML, CSS, and JavaScript in a single HTML file to create a simulation of the solar system. Pay extreme attention to the UI to make it as intuitive as possible. Ensure that every planet appears as a sphere and is labeled with its corresponding name.
example 1: o3 Mini , example 2: Deepseek R1
Q2 : https://huggingface.co./blog/prithivMLmods/o3-mini-vs-deepseek-r1#q2--web-solar-system-explorer
reacted to
jasoncorkill's
post with 🚀
3 days ago
Post
2570
We benchmarked
@xai-org
's Aurora model, as far as we know the first public evaluation of the model at scale.
We collected 401k human annotations in over the past ~2 days for this, we have uploaded all of the annotation data here on huggingface with a fully permissive license
Rapidata/xAI_Aurora_t2i_human_preferences
We collected 401k human annotations in over the past ~2 days for this, we have uploaded all of the annotation data here on huggingface with a fully permissive license
Rapidata/xAI_Aurora_t2i_human_preferences
reacted to
mkurman's
post with ❤️
3 days ago
Post
1901
Blurred-Thoughts Supervised Fine-Tuning (BT-SFT) 🤖
Can we teach a model to think completely on its own without reinforcement learning? Actually, yes.
We can do straightforward supervised fine-tuning using a relatively simple trick: blurring a part of CoT thoughts. But why is this effective?
We observed that various models differ in their thinking processes, and fine-tuning one model on another model’s thoughts (CoT) can sometimes be inefficient—often resulting in the model simply memorizing reasoning rather than learning how to actually think.
I discovered that this process can still be efficient if we clearly indicate when the model should start and stop thinking and uncover only a part of CoT and the expected answer, blurring the other part of CoT. This approach allows the model to learn only a portion of the thought process while still arriving at an expected answer.
To demonstrate this, you can watch my experimental BT-SFT on meditsolutions/Llama-3.2-SUN-2.5B-chat model, which was fine-tuned on 151 million tokens from the Magpie-Align/Magpie-Reasoning-V2-250K-CoT-Deepseek-R1-Llama-70B dataset.
Enjoy! 🚀
PS. If you were curious enough to read this, leave me a comment. It's always nice to chat with open-minded and intelligent ppl.
Can we teach a model to think completely on its own without reinforcement learning? Actually, yes.
We can do straightforward supervised fine-tuning using a relatively simple trick: blurring a part of CoT thoughts. But why is this effective?
We observed that various models differ in their thinking processes, and fine-tuning one model on another model’s thoughts (CoT) can sometimes be inefficient—often resulting in the model simply memorizing reasoning rather than learning how to actually think.
I discovered that this process can still be efficient if we clearly indicate when the model should start and stop thinking and uncover only a part of CoT and the expected answer, blurring the other part of CoT. This approach allows the model to learn only a portion of the thought process while still arriving at an expected answer.
To demonstrate this, you can watch my experimental BT-SFT on meditsolutions/Llama-3.2-SUN-2.5B-chat model, which was fine-tuned on 151 million tokens from the Magpie-Align/Magpie-Reasoning-V2-250K-CoT-Deepseek-R1-Llama-70B dataset.
Enjoy! 🚀
PS. If you were curious enough to read this, leave me a comment. It's always nice to chat with open-minded and intelligent ppl.
- 3 replies
reacted to
not-lain's
post with 🤗
4 days ago
Post
2537
I have just released a new blogpost about kv caching and its role in inference speedup 🚀
🔗 https://huggingface.co./blog/not-lain/kv-caching/
some takeaways :
🔗 https://huggingface.co./blog/not-lain/kv-caching/
some takeaways :
- 4 replies
reacted to
hexgrad's
post with 🚀🚀🚀🚀
4 days ago
Post
7399
hexgrad/Kokoro-82M got an upgrade! ⬆️ More voices, more languages,
GitHub: https://github.com/hexgrad/kokoro
PyPI: https://pypi.org/project/kokoro/
Space: hexgrad/Kokoro-TTS
pip install kokoro, and still 82M parameters.GitHub: https://github.com/hexgrad/kokoro
PyPI: https://pypi.org/project/kokoro/
Space: hexgrad/Kokoro-TTS
- 10 replies
replied to
their
post
5 days ago
I’m not sure, but if context continues to evolve in the future, the real OpenAI will improve more and more compared to paid AI providers. People will increasingly use open-source models for their domain-specific tasks.
posted
an
update
5 days ago
Post
4721
Deepswipe by
.
.
.
. Deepseek🐬🗿
Everything is now in recovery. 📉📈
.
.
.
. Deepseek🐬🗿
Everything is now in recovery. 📉📈
-
- 4 replies
reacted to
AdinaY's
post with 🔥
5 days ago
Post
2932
It’s not just a flood of model releases, papers are dropping just as fast 🚀
Here are the 10 most upvoted papers from the Chinese community:
👉 zh-ai-community/2025-january-papers-679933cbf0f3ced11f5a168a
Here are the 10 most upvoted papers from the Chinese community:
👉 zh-ai-community/2025-january-papers-679933cbf0f3ced11f5a168a
reacted to
m-ric's
post with 🔥
6 days ago
Post
3304
𝗧𝗵𝗲 𝗛𝘂𝗯 𝘄𝗲𝗹𝗰𝗼𝗺𝗲𝘀 𝗲𝘅𝘁𝗲𝗿𝗻𝗮𝗹 𝗶𝗻𝗳𝗲𝗿𝗲𝗻𝗰𝗲 𝗽𝗿𝗼𝘃𝗶𝗱𝗲𝗿𝘀!
✅ Hosting our own inference was not enough: now the Hub 4 new inference providers: fal, Replicate, SambaNova Systems, & Together AI.
Check model cards on the Hub: you can now, in 1 click, use inference from various providers (cf video demo)
Their inference can also be used through our Inference API client. There, you can use either your custom provider key, or your HF token, then billing will be handled directly on your HF account, as a way to centralize all expenses.
💸 Also, PRO users get 2$ inference credits per month!
Read more in the announcement 👉 https://huggingface.co./blog/inference-providers
✅ Hosting our own inference was not enough: now the Hub 4 new inference providers: fal, Replicate, SambaNova Systems, & Together AI.
Check model cards on the Hub: you can now, in 1 click, use inference from various providers (cf video demo)
Their inference can also be used through our Inference API client. There, you can use either your custom provider key, or your HF token, then billing will be handled directly on your HF account, as a way to centralize all expenses.
💸 Also, PRO users get 2$ inference credits per month!
Read more in the announcement 👉 https://huggingface.co./blog/inference-providers
- 1 reply
Incredible stat 📈
reacted to
victor's
post with 🚀
6 days ago
Post
2846
Finally, an open-source AI that turns your lyrics into full songs is here—meet YuE! Unlike other tools that only create short clips, YuE can make entire songs (up to 5 minutes) with vocals, melody, and instruments all working together. Letsss go!
m-a-p/YuE-s1-7B-anneal-en-cot
m-a-p/YuE-s1-7B-anneal-en-cot
- 1 reply
reacted to
clem's
post with ❤️
7 days ago
Post
6774
AI is not a zero-sum game. Open-source AI is the tide that lifts all boats!
reacted to
fdaudens's
post with ❤️
7 days ago
Post
7838
Yes, DeepSeek R1's release is impressive. But the real story is what happened in just 7 days after:
- Original release: 8 models, 540K downloads. Just the beginning...
- The community turned those open-weight models into +550 NEW models on Hugging Face. Total downloads? 2.5M—nearly 5X the originals.
The reason? DeepSeek models are open-weight, letting anyone build on top of them. Interesting to note that the community focused on quantized versions for better efficiency & accessibility. They want models that use less memory, run faster, and are more energy-efficient.
When you empower builders, innovation explodes. For everyone. 🚀
The most popular community model? @bartowski 's DeepSeek-R1-Distill-Qwen-32B-GGUF version — 1M downloads alone.
- Original release: 8 models, 540K downloads. Just the beginning...
- The community turned those open-weight models into +550 NEW models on Hugging Face. Total downloads? 2.5M—nearly 5X the originals.
The reason? DeepSeek models are open-weight, letting anyone build on top of them. Interesting to note that the community focused on quantized versions for better efficiency & accessibility. They want models that use less memory, run faster, and are more energy-efficient.
When you empower builders, innovation explodes. For everyone. 🚀
The most popular community model? @bartowski 's DeepSeek-R1-Distill-Qwen-32B-GGUF version — 1M downloads alone.
-
- 4 replies
reacted to
nicolay-r's
post with 🔥
7 days ago
Post
1748
📢 For those who wish to apply DeepSeek-R1 for handling tabular / streaming data using schema of prompts (CoT), the OpenRouter AI hosts API for accessing:
https://openrouter.ai/deepseek/deepseek-r1
The no-string option to quick start with using DeepSeek-R1 includes three steps:
✅ OpenRouter provider: https://github.com/nicolay-r/nlp-thirdgate/blob/master/llm/open_router.py
✅ Bulk-chain for infering data: https://github.com/nicolay-r/bulk-chain
✅ Json Schema for Chain-of-Though reasoning (see screenshot 📷 below)
📺 below is a screenshot of how to quick start the demo, in which you can test your schema for LLM responses. It would ask to type all the parameters first for completing the requests (which is
📃 To apply it for JSONL/CSV data, you can use
⏳ As for time, OpenRouter finds me relatively slow with 30~40 seconds per request
Models:
deepseek-ai/DeepSeek-R1
https://openrouter.ai/deepseek/deepseek-r1
The no-string option to quick start with using DeepSeek-R1 includes three steps:
✅ OpenRouter provider: https://github.com/nicolay-r/nlp-thirdgate/blob/master/llm/open_router.py
✅ Bulk-chain for infering data: https://github.com/nicolay-r/bulk-chain
✅ Json Schema for Chain-of-Though reasoning (see screenshot 📷 below)
📺 below is a screenshot of how to quick start the demo, in which you can test your schema for LLM responses. It would ask to type all the parameters first for completing the requests (which is
text within this example).📃 To apply it for JSONL/CSV data, you can use
--src shell parameter for passing the related file⏳ As for time, OpenRouter finds me relatively slow with 30~40 seconds per request
Models:
deepseek-ai/DeepSeek-R1
-
- 1 reply
reacted to
AdinaY's
post with 🔥
9 days ago
Post
1345
Baichuan is making big moves today 🔥
✨ Launched All-Scenario Reasoning Model (language, visual, and search reasoning capabilities) , with medical expertise as one of its key highlights.
https://ying.baichuan-ai.com/chat
✨ Released Baichuan-M1-14B Medical LLM on the hub
Available in both Base and Instruct versions, support English & Chinese.
Model:
baichuan-inc/Baichuan-M1-14B-Base
baichuan-inc/Baichuan-M1-14B-Instruct
✨ Launched All-Scenario Reasoning Model (language, visual, and search reasoning capabilities) , with medical expertise as one of its key highlights.
https://ying.baichuan-ai.com/chat
✨ Released Baichuan-M1-14B Medical LLM on the hub
Available in both Base and Instruct versions, support English & Chinese.
Model:
baichuan-inc/Baichuan-M1-14B-Base
baichuan-inc/Baichuan-M1-14B-Instruct
reacted to
burtenshaw's
post with 🤯
10 days ago
Post
2236
AI was built on side projects!
reacted to
AdinaY's
post with 🔥
11 days ago
Post
1410
VideoLLaMA 3🔥multimodal foundation models for Image and Video Understanding by DAMO Alibaba
Model: DAMO-NLP-SG/videollama3-678cdda9281a0e32fe79af15
Paper: VideoLLaMA 3: Frontier Multimodal Foundation Models for Image and Video Understanding (2501.13106)
✨ 2B/7B
✨ Apache2.0
Model: DAMO-NLP-SG/videollama3-678cdda9281a0e32fe79af15
Paper: VideoLLaMA 3: Frontier Multimodal Foundation Models for Image and Video Understanding (2501.13106)
✨ 2B/7B
✨ Apache2.0
- 1 reply