| Hey, | Welcome to AlphaSignal – the most read newsletter by AI developers.
We bring you the top 1% of news, papers, models, and repos, all summarized to keep you updated on the latest in AI. |
|
|
|
IN TODAY'S SIGNAL
| Read time: 4 min 57 sec | 🎖️ Top News
📌 Assembly AI
⚡️ Trending Signals
💻 Top Tutorials
🧠 How To |
|
|
|
If you're enjoying AlphaSignal please forward this email to a colleague.
It helps us keep this content free. |
|
|
|
TOP NEWS
| Long-Context LLM |
|
⇧ 4,103 Likes |
| What's New |
Alibaba’s Qwen team launches Qwen2.5-1M, a set of open-source models that support a context length of up to 1 million tokens. These models, Qwen2.5-7B-Instruct-1M and Qwen2.5-14B-Instruct-1M, mark the first in the Qwen series to handle such long context inputs. The release builds on Qwen2.5-Turbo, which supports a 128k token context.
Key performance highlights - 7B model performs similarly to specialized long-context models.
- 14B model exceeds performance of both Qwen2.5-Turbo and GPT-4o-mini.
- Supports up to 1 million tokens in context, a significant increase from the previous 128k token limit.
- Uses sparse attention and Dual Chunk Attention (DCA) to manage large input sequences effectively.
Technical Details The core innovation lies in the inference framework based on vLLM, utilizing sparse attention. This design boosts performance by processing inputs 7x faster and optimizing memory usage. The models deliver efficient results while managing long-context sequences.
The training process involved gradual context length increases, beginning at 4k tokens and progressing to longer sequences. This approach enabled the models to handle both short and long-context tasks efficiently.
Access and Availability
You can experience Qwen2.5-1M models through various platforms: - Qwen Chat: Play with the Qwen2.5-Turbo supporting 1M tokens directly in the Qwen Chat.
- Hugging Face: Explore the models on Hugging Face.
- Modelscope: Access the models on Modelscope.
| |
| Try now
|
|
|
|
| Improve Retention with Accurate Transcriptions | Use AssemblyAI’s Speech-to-Text API to process conversations and extract insights fast.
- Get accurate transcriptions that reduce churn and improve decision-making.
- Access tools like Speaker Diarization, Summarization, and Language Detection in one API.
- Integrate in minutes with minimal setup using a developer-friendly interface.
Join over 200,000 developers already using AssemblyAI to build reliable, voice-driven tools. |
| Start now
| partner with us → |
|
|
|
TRENDING SIGNALS
| AI Tool |
|
⇧ 718 Likes | |
AI Model |
|
⇧ 8,394 Likes | |
AI Assistant |
|
⇧ 9,264 Likes | |
Multimodal AI |
|
⇧ 493 Likes | |
AI Assistant |
|
⇧ 8,494 Likes | | |
|
|
|
|
TOP TUTORIALS
| Reasoning Model |
| ⇧ 2,483 Likes |
Learn how to use DeepSeek-R1, an open-source reasoning model with a 14B distilled version. This tutorial reviews its training methods, explains downloading the model via Ollama, and demonstrates JSON-mode testing. Test its local "deep research" assistant, which performs web research and iterative summarization with reflection for improved results. |
| Vision and Agent Interaction |
| ⇧ 854 Likes |
In this blog, learn how to integrate vision capabilities into autonomous agents using smolagents. This tutorial explains passing images to agents in two ways: at initialization or dynamically via callbacks. It demonstrates building a web-browsing agent with vision using the MultiStepAgent class and helium. The agent performs actions like navigation, popup handling, and dynamic webpage analysis. |
| LLM Evaluation |
| ⇧ 206 Likes |
Learn how to evaluate LLM-generated summaries using a practical, research-backed, quantitative framework. This guide explores challenges in summarization evaluation, defines key quality metrics (conciseness, coherence), and improves the Summarization Metric in the DeepEval framework. Includes a GitHub notebook for applying these methods to assess summaries of long-form content systematically. |
| |
| |
|
|
HOW TO
| How to Run DeepSeek R1 Locally | This guide walks you through setting up and running the DeepSeek R1 model on your machine using Ollama and Chatbox.
Here is how to set up DeepSeek R1:
Step 1: Install Ollama
Download Ollama, a tool for running AI models locally, from https://ollama.com/download.
Step 2: Run the DeepSeek R1 model
Open your terminal and run one of the following commands based on your hardware: - Smallest (1.5B): ollama run deepseek-r1:1.5b
- Medium (8B): ollama run deepseek-r1:8b
- Largest (70B, requires a powerful GPU): ollama run deepseek-r1:70b
Step 3: Install Chatbox
Download Chatbox, a free desktop interface, from https://chatboxai.app.
Step 4: Configure Chatbox
- Open settings and switch the model provider to "Ollama."
- Set the API host to http://127.0.0.1:11434.
Step 5: Start chatting
Select DeepSeek R1 in Chatbox, and you’re ready to go! |
|
|
|
At Alpha Signal, our mission is to build a sharp, engaged community focused on AI, machine learning, and cutting-edge language models, helping over 200,000 developers stay informed and ahead. We’re passionate about curating the best in AI, from top research and trending technical blogs to expert insights and tailored job opportunities. We keep you connected to the breakthroughs and discussions that matter, so you can stay in the loop without endless searching. We also work closely with partners who value the future of AI, including employers and advertisers who want to reach an audience as passionate about AI as we are.
Our partnerships are based on shared values of ethics, responsibility, and a commitment to building a better world through technology.Privacy is a priority at Alpha Signal. Our Privacy Policy clearly explains how we collect, store, and use your personal and non-personal information. By using our website, you accept these terms, which you can review on our website. This policy applies across all Alpha Signal pages, outlining your rights and how to contact us if you want to adjust the use of your information. We’re based in the United States. By using our site, you agree to be governed by U.S. laws. |
|
|
|
Looking to promote your company, product, service, or event to 200,000+ AI developers? Let's work together.
|
| Work with us
|
|
|
|
|