Hugging Face's logo
Djuunaa's picture
130 12 317

Djuunaa

djuna
drilbo's profile picture ltim's profile picture Triangle104's profile picture
·

AI & ML interests

None yet

Recent Activity

reacted to davidberenstein1957's post with 👀 1 minute ago
Let's uncover the post-training dataset from DeepSeek-R1 with Magpie! Pass pre-query tokens `<｜begin▁of▁sentence｜>User: `, let the model generate the rest. We can get realistic examples! Gist: https://gist.github.com/davidberenstein1957/3f20046ce57395a6aba13f8b4e956b59
liked a model about 4 hours ago
Alibaba-NLP/gte-modernbert-base
reacted to hbseong's post with 👀 about 5 hours ago
🚨🔥 New Release Alert! 🔥🚨 Introducing the 435M model that outperforms Llama-Guard-3-8B while slashing 75% of the computation cost! 💻💥 👉 Check it out: https://huggingface.co/hbseong/HarmAug-Guard (Yes, INFERENCE CODE INCLUDED! 💡) More details in our paper: https://arxiv.org/abs/2410.01524 📜 #HarmAug #LLM # Safety #EfficiencyBoost #Research #AI #MachineLearning
View all activity

Organizations

Djuna Test Lab's profile picture

djuna's activity

reacted to davidberenstein1957's post with 👀 1 minute ago
liked a model about 4 hours ago
reacted to hbseong's post with 👀 about 5 hours ago
view post
Post
925
🚨🔥 New Release Alert! 🔥🚨

Introducing the 435M model that outperforms Llama-Guard-3-8B while slashing 75% of the computation cost! 💻💥
👉 Check it out: hbseong/HarmAug-Guard (Yes, INFERENCE CODE INCLUDED! 💡)

More details in our paper: https://arxiv.org/abs/2410.01524 📜

#HarmAug #LLM # Safety #EfficiencyBoost #Research #AI #MachineLearning
  • 1 reply
·
reacted to lewtun's post with 🔥 about 6 hours ago
view post
Post
4168
We are reproducing the full DeepSeek R1 data and training pipeline so everybody can use their recipe. Instead of doing it in secret we can do it together in the open!

🧪 Step 1: replicate the R1-Distill models by distilling a high-quality reasoning corpus from DeepSeek-R1.

🧠 Step 2: replicate the pure RL pipeline that DeepSeek used to create R1-Zero. This will involve curating new, large-scale datasets for math, reasoning, and code.

🔥 Step 3: show we can go from base model -> SFT -> RL via multi-stage training.

Follow along: https://github.com/huggingface/open-r1
  • 1 reply
·
liked 2 models about 10 hours ago
liked a model about 11 hours ago
liked a dataset about 12 hours ago
liked a model about 13 hours ago
New activity in djuna/MN-Chinofun-12B-4 about 13 hours ago

Adding Evaluation Results
#1 opened about 13 hours ago by
djuna
updated a model about 13 hours ago
updated a collection about 13 hours ago
liked a model about 13 hours ago
liked 3 models about 16 hours ago
New activity in arcee-ai/mergekit-gui about 16 hours ago

Error: Unimplemented merge method sce

5
#35 opened 3 days ago by
xi0v
upvoted a collection about 16 hours ago
reacted to onekq's post with 🔥 about 16 hours ago
view post
Post
1848
So 🐋DeepSeek🐋 hits the mainstream media. But it has been a star in our little cult for at least 6 months. Its meteoric success is not overnight, but two years in the making.

To learn their history, just look at their 🤗 repo https://huggingface.co/deepseek-ai

* End of 2023, they launched the first model (pretrained by themselves) following Llama 2 architecture
* June 2024, v2 (MoE architecture) surpassed Gemini 1.5, but behind Mistral
* September, v2.5 surpassed GPT 4o mini
* December, v3 surpassed GPT 4o
* Now R1 surpassed o1

Most importantly, if you think DeepSeek success is singular and unrivaled, that's WRONG. The following models are also near or equal the o1 bar.

* Minimax-01
* Kimi k1.5
* Doubao 1.5 pro