Di Zhang
qq8933
AI & ML interests
AI4Chem, LLM, Green LLM
Recent Activity
posted
an
update
6 days ago
News! ChemVLM Codes Opensource Now! https://github.com/AI4Chem/ChemVlm
authored
a paper
12 days ago
StyleMaster: Stylize Your Video with Artistic Generation and Translation
authored
a paper
12 days ago
SynCamMaster: Synchronizing Multi-Camera Video Generation from Diverse
Viewpoints
Organizations
qq8933's activity
posted
an
update
6 days ago
replied to
their
post
14 days ago
not perfect, but just works:)
posted
an
update
14 days ago
Post
2517
LLaMA-O1-PRM and LLaMA-O1-Reinforcement will release in this weekend.
We have implemented a novel Reinforcement finetune(RFT) pipeline that taught models learning reasoning and reward labeling without human annotation.
We have implemented a novel Reinforcement finetune(RFT) pipeline that taught models learning reasoning and reward labeling without human annotation.
-
- 3 replies
posted
an
update
15 days ago
Post
1649
ChemVLM has been accepted by AAAI2025!
Seeing and Understanding: Bridging Vision with Chemical Knowledge Via ChemVLM (2408.07246)
Try have a chat wiht him๐ค.
AI4Chem/ChemVLM-26B-1-2
Seeing and Understanding: Bridging Vision with Chemical Knowledge Via ChemVLM (2408.07246)
Try have a chat wiht him๐ค.
AI4Chem/ChemVLM-26B-1-2
replied to
their
post
23 days ago
We will write a short technical report for current progress.
reacted to
their
post with ๐
23 days ago
Post
3037
The first version of LLaMA-O1 has been uploaded to HF now!Here We Come!
Supervised:
SimpleBerry/LLaMA-O1-Supervised-1129
Base(Pretrain):
SimpleBerry/LLaMA-O1-Base-1127
Supervised Finetune Dataset:
SimpleBerry/OpenLongCoT-SFT
Pretraining Dataset:
SimpleBerry/OpenLongCoT-Pretrain-1202
RLHF is on the way! View our GitHub Repo:
https://github.com/SimpleBerry/LLaMA-O1
Our ongoing related researches:
Accessing GPT-4 level Mathematical Olympiad Solutions via Monte Carlo Tree Self-refine with LLaMa-3 8B (2406.07394)
LLaMA-Berry: Pairwise Optimization for O1-like Olympiad-Level Mathematical Reasoning (2410.02884)
Critic-V: VLM Critics Help Catch VLM Errors in Multimodal Reasoning (2411.18203)
@AdinaY @akhaliq @jwu323
------
GGUF:https://huggingface.co/Lyte/LLaMA-O1-Supervised-1129-Q4_K_M-GGUF
online Demo (CPU-only): SimpleBerry/LLaMA-O1-Supervised-1129-Demo
Supervised:
SimpleBerry/LLaMA-O1-Supervised-1129
Base(Pretrain):
SimpleBerry/LLaMA-O1-Base-1127
Supervised Finetune Dataset:
SimpleBerry/OpenLongCoT-SFT
Pretraining Dataset:
SimpleBerry/OpenLongCoT-Pretrain-1202
RLHF is on the way! View our GitHub Repo:
https://github.com/SimpleBerry/LLaMA-O1
Our ongoing related researches:
Accessing GPT-4 level Mathematical Olympiad Solutions via Monte Carlo Tree Self-refine with LLaMa-3 8B (2406.07394)
LLaMA-Berry: Pairwise Optimization for O1-like Olympiad-Level Mathematical Reasoning (2410.02884)
Critic-V: VLM Critics Help Catch VLM Errors in Multimodal Reasoning (2411.18203)
@AdinaY @akhaliq @jwu323
------
GGUF:https://huggingface.co/Lyte/LLaMA-O1-Supervised-1129-Q4_K_M-GGUF
online Demo (CPU-only): SimpleBerry/LLaMA-O1-Supervised-1129-Demo
-
- 3 replies
replied to
their
post
23 days ago
This comment has been hidden
posted
an
update
23 days ago
Post
3037
The first version of LLaMA-O1 has been uploaded to HF now!Here We Come!
Supervised:
SimpleBerry/LLaMA-O1-Supervised-1129
Base(Pretrain):
SimpleBerry/LLaMA-O1-Base-1127
Supervised Finetune Dataset:
SimpleBerry/OpenLongCoT-SFT
Pretraining Dataset:
SimpleBerry/OpenLongCoT-Pretrain-1202
RLHF is on the way! View our GitHub Repo:
https://github.com/SimpleBerry/LLaMA-O1
Our ongoing related researches:
Accessing GPT-4 level Mathematical Olympiad Solutions via Monte Carlo Tree Self-refine with LLaMa-3 8B (2406.07394)
LLaMA-Berry: Pairwise Optimization for O1-like Olympiad-Level Mathematical Reasoning (2410.02884)
Critic-V: VLM Critics Help Catch VLM Errors in Multimodal Reasoning (2411.18203)
@AdinaY @akhaliq @jwu323
------
GGUF:https://huggingface.co/Lyte/LLaMA-O1-Supervised-1129-Q4_K_M-GGUF
online Demo (CPU-only): SimpleBerry/LLaMA-O1-Supervised-1129-Demo
Supervised:
SimpleBerry/LLaMA-O1-Supervised-1129
Base(Pretrain):
SimpleBerry/LLaMA-O1-Base-1127
Supervised Finetune Dataset:
SimpleBerry/OpenLongCoT-SFT
Pretraining Dataset:
SimpleBerry/OpenLongCoT-Pretrain-1202
RLHF is on the way! View our GitHub Repo:
https://github.com/SimpleBerry/LLaMA-O1
Our ongoing related researches:
Accessing GPT-4 level Mathematical Olympiad Solutions via Monte Carlo Tree Self-refine with LLaMa-3 8B (2406.07394)
LLaMA-Berry: Pairwise Optimization for O1-like Olympiad-Level Mathematical Reasoning (2410.02884)
Critic-V: VLM Critics Help Catch VLM Errors in Multimodal Reasoning (2411.18203)
@AdinaY @akhaliq @jwu323
------
GGUF:https://huggingface.co/Lyte/LLaMA-O1-Supervised-1129-Q4_K_M-GGUF
online Demo (CPU-only): SimpleBerry/LLaMA-O1-Supervised-1129-Demo
-
-
- 3 replies
posted
an
update
26 days ago
Post
1343
LLaMA-O1 Base and SFT model will be uploaded to HF today.
RLHF pipeline already ready, still waiting for data sampling.
RLHF pipeline already ready, still waiting for data sampling.
-
- 1 reply
Stay Tuned!
reacted to
jwu323's
post with ๐
28 days ago
Post
1351
We are excited to announce a new internal project, Rome, focused on advancing LLM reasoning. The code and accompanying paper will be released soon. Stay tuned!
-
- 2 replies
replied to
their
post
about 1 month ago
You're Genius!
replied to
their
post
about 2 months ago
main.py is the entry for finetune, but codes need further improvements, see 'Call for contributors'
posted
an
update
about 2 months ago
Post
2406
Discovered an outrageous bug on the ChatGPT official website, especially for those using ad-blocking plugins. Please make sure to add
For users with Chinese IP addresses, consider adding this URL to the rules of your U.S. node, as the response headers from this site will report the user's physical location to GPT.
browser-intake-datadoghq.com to your ad block whitelist. The ChatGPT webpage relies on this site for heartbeat detection, but since it belongs to an ad tracking network, it's included in major ad-blocking lists. (If you're using Clash, also remember to add it to the whitelist.) Failing to do so may cause the ChatGPT web interface to display a greyed-out send button after clicking, with no response.For users with Chinese IP addresses, consider adding this URL to the rules of your U.S. node, as the response headers from this site will report the user's physical location to GPT.
-
- 3 replies
posted
an
update
about 2 months ago
Post
6248
LLaMA-O1: Open Large Reasoning Model Frameworks For Training, Inference and Evaluation With PyTorch and HuggingFace
Large Reasoning Models powered by Monte Carlo Tree Search (MCTS), Self-Play Reinforcement Learning, PPO, AlphaGo Zero's dua policy paradigm and Large Language Models!
https://github.com/SimpleBerry/LLaMA-O1/
What will happen when you compound MCTS โค LLM โค Self-Play โคRLHF?
Just a little bite of strawberry!๐
Past related works:
LLaMA-Berry: Pairwise Optimization for O1-like Olympiad-Level Mathematical Reasoning (2410.02884)
Accessing GPT-4 level Mathematical Olympiad Solutions via Monte Carlo Tree Self-refine with LLaMa-3 8B (2406.07394)
Large Reasoning Models powered by Monte Carlo Tree Search (MCTS), Self-Play Reinforcement Learning, PPO, AlphaGo Zero's dua policy paradigm and Large Language Models!
https://github.com/SimpleBerry/LLaMA-O1/
What will happen when you compound MCTS โค LLM โค Self-Play โคRLHF?
Just a little bite of strawberry!๐
Past related works:
LLaMA-Berry: Pairwise Optimization for O1-like Olympiad-Level Mathematical Reasoning (2410.02884)
Accessing GPT-4 level Mathematical Olympiad Solutions via Monte Carlo Tree Self-refine with LLaMa-3 8B (2406.07394)
-
- 2 replies
posted
an
update
4 months ago
Post
1537
๐ Introducing ChemVLM, the first open-source multimodal large language model dedicated to chemistry!
๐Comparable performances with commercial models or specific OCR model but with dialogue capabilities!
โจ2B/26B Models Here! AI4Chem/ChemVLM-26B
Seeing and Understanding: Bridging Vision with Chemical Knowledge Via ChemVLM (2408.07246)
๐Comparable performances with commercial models or specific OCR model but with dialogue capabilities!
โจ2B/26B Models Here! AI4Chem/ChemVLM-26B
Seeing and Understanding: Bridging Vision with Chemical Knowledge Via ChemVLM (2408.07246)
replied to
their
post
5 months ago
ๅ็ซฏๅผๅธธ๏ผๆๆไบ๏ผๅจไฟฎๅค
posted
an
update
6 months ago
Post
654
Preview:
We will open source the 2.5B ChemVLM and the tool-enhanced ChemLLM-7B in the near future
We will open source the 2.5B ChemVLM and the tool-enhanced ChemLLM-7B in the near future
posted
an
update
6 months ago
Post
737
A great work based on ChemLLM from Open-source community!
Automatic Scientific Discovery guided by LLM!
https://github.com/zyzisastudyreallyhardguy/LLM4SD
Automatic Scientific Discovery guided by LLM!
https://github.com/zyzisastudyreallyhardguy/LLM4SD
reacted to
their
post with ๐
6 months ago
Post
2617
Tools Ready!
Thanks to ChemCrow's great work, ChemLLM supports proficiency toolkits Now, Include,
Molecule Name Retrivel
Molecule Property Query
Patent Check
Molecule Safety Query
Try it on chemllm.org
Thanks to ChemCrow's great work, ChemLLM supports proficiency toolkits Now, Include,
Molecule Name Retrivel
Molecule Property Query
Patent Check
Molecule Safety Query
Try it on chemllm.org
-
- 2 replies