Does Daily Software Engineering Work Need Reasoning Models?
β’
5
Yi Cui PRO
onekq
AI & ML interests
Benchmark, Code Generation Model
Recent Activity
updated
a Space
about 1 month ago
onekq-ai/WebApp1K-models-leaderboard
posted
an
update
about 2 months ago
October version of Claude 3.5 lifts SOTA (set by its June version) by 7 points.
https://huggingface.co/spaces/onekq-ai/WebApp1K-models-leaderboard
Closed sourced models are widening the gap again.
Note: Our frontier leaderboard now uses double test scenarios because the single-scenario test suit has been saturated.
new activity
about 2 months ago
onekq-ai/WebApp1K-models-leaderboard:All the clickable links are not accessible...
Articles
Organizations
onekq's activity
posted
an
update
about 2 months ago
Post
565
October version of Claude 3.5 lifts SOTA (set by its June version) by 7 points.
onekq-ai/WebApp1K-models-leaderboard
Closed sourced models are widening the gap again.
Note: Our frontier leaderboard now uses double test scenarios because the single-scenario test suit has been saturated.
onekq-ai/WebApp1K-models-leaderboard
Closed sourced models are widening the gap again.
Note: Our frontier leaderboard now uses double test scenarios because the single-scenario test suit has been saturated.
posted
an
update
2 months ago
Post
1846
I'm now working on finetuning of coding models. If you are GPU-hungry like me, you will find quantized models very helpful. But quantization for finetuning and inference are different and incompatible. So I made two collections here.
Inference (GGUF, via Ollama, CPU is enough)
onekq-ai/ollama-ready-coding-models-67118c3cfa1af2cf04a926d6
Finetuning (Bitsandbytes, QLora, GPU is needed)
onekq-ai/qlora-ready-coding-models-67118771ce001b8f4cf946b2
For quantization, the inference models are far more popular on HF than finetuning models. I use https://huggingface.co/QuantFactory to generate inference models (GGUF), and there are a few other choices.
But there hasn't been such a service for finetuning models. DIY isn't too hard though. I made a few myself and you can find the script in the model cards. If the original model is small enough, you can even do it on a free T4 (available via Google Colab).
If you know a (small) coding model worthy of quantization, please let me know and I'd love to add it to the collections.
Inference (GGUF, via Ollama, CPU is enough)
onekq-ai/ollama-ready-coding-models-67118c3cfa1af2cf04a926d6
Finetuning (Bitsandbytes, QLora, GPU is needed)
onekq-ai/qlora-ready-coding-models-67118771ce001b8f4cf946b2
For quantization, the inference models are far more popular on HF than finetuning models. I use https://huggingface.co/QuantFactory to generate inference models (GGUF), and there are a few other choices.
But there hasn't been such a service for finetuning models. DIY isn't too hard though. I made a few myself and you can find the script in the model cards. If the original model is small enough, you can even do it on a free T4 (available via Google Colab).
If you know a (small) coding model worthy of quantization, please let me know and I'd love to add it to the collections.
reacted to
fdaudens's
post with π₯
3 months ago
Post
1792
Exciting news in AI: Molmo, a groundbreaking family of open-source multimodal models, has just been announced! π
Key points:
- Closes the gap with proprietary systems on benchmarks & human evals
- Trained on high-quality data (< 1M image-text pairs vs billions)
- Introduces pointing capability for rich interactions
- Fully open weights, data, and training code
The 72B model outperforms several proprietary systems, while the 1B model nearly matches GPT-4V. Small is indeed the new big in AI!
There's an interactive demo available using Molmo-7B-D. Definitely worth checking out to see its capabilities firsthand.
All model weights, data, and code will be released soon. This is a significant step towards truly open, cutting-edge multimodal AI.
The future of AI research and applications is looking brighter than ever! π€πΌοΈ
π Demo: https://molmo.allenai.org/
π Models: allenai/molmo-66f379e6fe3b8ef090a8ca19
#AI #MachineLearning #OpenSource #ComputerVision
Key points:
- Closes the gap with proprietary systems on benchmarks & human evals
- Trained on high-quality data (< 1M image-text pairs vs billions)
- Introduces pointing capability for rich interactions
- Fully open weights, data, and training code
The 72B model outperforms several proprietary systems, while the 1B model nearly matches GPT-4V. Small is indeed the new big in AI!
There's an interactive demo available using Molmo-7B-D. Definitely worth checking out to see its capabilities firsthand.
All model weights, data, and code will be released soon. This is a significant step towards truly open, cutting-edge multimodal AI.
The future of AI research and applications is looking brighter than ever! π€πΌοΈ
π Demo: https://molmo.allenai.org/
π Models: allenai/molmo-66f379e6fe3b8ef090a8ca19
#AI #MachineLearning #OpenSource #ComputerVision
reacted to
victor's
post with ππ€
3 months ago
Post
5552
π Calling all Hugging Face users! We want to hear from YOU!
What feature or improvement would make the biggest impact on Hugging Face?
Whether it's the Hub, better documentation, new integrations, or something completely different β we're all ears!
Your feedback shapes the future of Hugging Face. Drop your ideas in the comments below! π
What feature or improvement would make the biggest impact on Hugging Face?
Whether it's the Hub, better documentation, new integrations, or something completely different β we're all ears!
Your feedback shapes the future of Hugging Face. Drop your ideas in the comments below! π
- 168 replies
reacted to
their
post with π§
3 months ago
Post
2556
Here is my latest study on OpenAIπo1π.
A Case Study of Web App Coding with OpenAI Reasoning Models (2409.13773)
I wrote an easy-to-read blogpost to explain finding.
https://huggingface.co/blog/onekq/daily-software-engineering-work-reasoning-models
INSTRUCTION FOLLOWING is the key.
100% instruction following + Reasoning = new SOTA
But if the model misses or misunderstands one instruction, it can perform far worse than non-reasoning models.
A Case Study of Web App Coding with OpenAI Reasoning Models (2409.13773)
I wrote an easy-to-read blogpost to explain finding.
https://huggingface.co/blog/onekq/daily-software-engineering-work-reasoning-models
INSTRUCTION FOLLOWING is the key.
100% instruction following + Reasoning = new SOTA
But if the model misses or misunderstands one instruction, it can perform far worse than non-reasoning models.
posted
an
update
3 months ago
Post
2556
Here is my latest study on OpenAIπo1π.
A Case Study of Web App Coding with OpenAI Reasoning Models (2409.13773)
I wrote an easy-to-read blogpost to explain finding.
https://huggingface.co/blog/onekq/daily-software-engineering-work-reasoning-models
INSTRUCTION FOLLOWING is the key.
100% instruction following + Reasoning = new SOTA
But if the model misses or misunderstands one instruction, it can perform far worse than non-reasoning models.
A Case Study of Web App Coding with OpenAI Reasoning Models (2409.13773)
I wrote an easy-to-read blogpost to explain finding.
https://huggingface.co/blog/onekq/daily-software-engineering-work-reasoning-models
INSTRUCTION FOLLOWING is the key.
100% instruction following + Reasoning = new SOTA
But if the model misses or misunderstands one instruction, it can perform far worse than non-reasoning models.
posted
an
update
3 months ago
Post
424
Announce π WebApp1K-Duo π
onekq-ai/WebApp1K-Duo-React
This is to keep up the challenge after OpenAI o1 models saturated the WebApp1K benchmark. The new benchmark brings SOTA to 67%. Let the hill climbing commence!
onekq-ai/WebApp1K-models-leaderboard
PS: I will publish more findings soon.
onekq-ai/WebApp1K-Duo-React
This is to keep up the challenge after OpenAI o1 models saturated the WebApp1K benchmark. The new benchmark brings SOTA to 67%. Let the hill climbing commence!
onekq-ai/WebApp1K-models-leaderboard
PS: I will publish more findings soon.
reacted to
KingNish's
post with π
3 months ago
Post
3573
Mistral Nemo is better than many models in 1st grader level reasoning.
replied to
zhabotorabi's
post
3 months ago
the Mistral API? the model name is probably diffrent. I used mistral-large-2 but had to use the name mistral-large-latest. The team will help you via chat.
posted
an
update
3 months ago
Post
549
π DeepSeek π2.5 is hands-down the best open-source model, leaving its peers way behind. It even beats GPT-4o mini.
onekq-ai/WebApp1K-models-leaderboard
The inference of the official API is painfully slow though. I heard the team is short on GPUs (well, who isn't).
onekq-ai/WebApp1K-models-leaderboard
The inference of the official API is painfully slow though. I heard the team is short on GPUs (well, who isn't).
reacted to
aaditya's
post with π
3 months ago
Post
2557
Last Week in Medical AI: Top Research Papers/Models
π (September 7 - September 14, 2024)
π Medical AI Paper of the week
Chai-1 Foundation model molecular structure prediction
Medical LLMs & Benchmarks
- BrainWave: A Brain Signal Foundation Model
- DS-ViT: Vision Transformer for Alzheimerβs Diagnosis
- EyeCLIP: Visualβlanguage model for ophthalmic
- Segment Anything Model for Tumor Segmentation
- MEDIC: Evaluating LLMs in Clinical Applications
Medical LLM Applications
- KARGEN: Radiology Report Generation LLMs
- DrugAgent: Explainable Drug Repurposing Agents
- Improving RAG in Medicine with Follow-up Questions
Frameworks and Methodologies
- Infrastructure for Automatic Cell Segmentation
- Data Alignment for Dermatology AI
- Diagnostic Reasoning in Natural Language
- Two-Stage Instruction Fine-tuning Approach for Med
AI in Healthcare Ethics
- Concerns and Choices of Using LLMs for Healthcare
- Understanding Fairness in Recommender Systems
- Towards Fairer Health Recommendations
Check the full thread: https://x.com/OpenlifesciAI/status/1832476252260712788
Thank you for your continued support and love for this series! Stay up-to-date with weekly updates on Medical LLMs, datasets, and top research papers by following @aaditya π€
π (September 7 - September 14, 2024)
π Medical AI Paper of the week
Chai-1 Foundation model molecular structure prediction
Medical LLMs & Benchmarks
- BrainWave: A Brain Signal Foundation Model
- DS-ViT: Vision Transformer for Alzheimerβs Diagnosis
- EyeCLIP: Visualβlanguage model for ophthalmic
- Segment Anything Model for Tumor Segmentation
- MEDIC: Evaluating LLMs in Clinical Applications
Medical LLM Applications
- KARGEN: Radiology Report Generation LLMs
- DrugAgent: Explainable Drug Repurposing Agents
- Improving RAG in Medicine with Follow-up Questions
Frameworks and Methodologies
- Infrastructure for Automatic Cell Segmentation
- Data Alignment for Dermatology AI
- Diagnostic Reasoning in Natural Language
- Two-Stage Instruction Fine-tuning Approach for Med
AI in Healthcare Ethics
- Concerns and Choices of Using LLMs for Healthcare
- Understanding Fairness in Recommender Systems
- Towards Fairer Health Recommendations
Check the full thread: https://x.com/OpenlifesciAI/status/1832476252260712788
Thank you for your continued support and love for this series! Stay up-to-date with weekly updates on Medical LLMs, datasets, and top research papers by following @aaditya π€
replied to
their
post
3 months ago
pass@1 for πo1-miniπ: 0.94!!
πΈπΈπΈπΈ
#gpt #o1 #inference #RL #selfplay #WebApp1K
posted
an
update
3 months ago
Post
1121
If your plan keeps changing it's a sign that you are living the moment.
I just got the pass@1 result of GPT πo1-previewπ : 0.95!!!
This means my benchmark is cast into oblivion, I need to up the ante. I am all ears to suggestions. onekq-ai/WebApp1K-models-leaderboard
I just got the pass@1 result of GPT πo1-previewπ : 0.95!!!
This means my benchmark is cast into oblivion, I need to up the ante. I am all ears to suggestions. onekq-ai/WebApp1K-models-leaderboard
-
- 1 reply