IN TODAY'S SIGNAL |
|
🎖️ Top News
📌 WorkOS
⚡️ Top 5 Signals
📚 DeepAtlas
🛠️ Top Repos
-
Tokencost: calculate the cost of using major LLM APIs efficiently.
-
ComfyUI: powerful, modular stable diffusion GUI with graph-based interface
-
Awesome-notebooks: extensive AI notebook templates for various use cases.
🧠 Pytorch Tip
|
|
Read Time: 4 min 20 sec |
|
|
|
|
Enjoying this newsletter?
Please forward it to a friend or colleague. It helps us keep this content free. |
|
|
|
TOP NEWS |
|
Language Models |
Anthropic releases Claude 3.5, an LLM that surpasses GPT-4o while being 2x faster and 80% cheaper than Claude 3 Opus |
|
⇧ 4421 Likes |
 |
What's New |
|
Anthropic has introduced Claude 3.5 Sonnet, an advanced AI model that offers significant improvements in speed, cost efficiency, and performance over its predecessor, Claude 3 Opus.
The model costs $3 per million input tokens and $15 per million output tokens, with a 200K token context window.
Key Performance Enhancements
- Claude 3.5 Sonnet is twice as fast and 80% cheaper than Claude 3 Opus.
- The model excels in graduate-level reasoning (GPQA), undergraduate-level knowledge (MMLU), and coding proficiency (HumanEval).
Advanced Problem-Solving in Coding
- Claude 3.5 Sonnet has demonstrated a success rate of 64% on agentic coding problems, compared to the 38% success rate of Claude 3 Opus.
- These capabilities make it highly effective for tasks requiring advanced reasoning and the ability to interpret complex instructions.
Enhanced Vision Capabilities
- The model sets new benchmarks in vision capabilities, significantly outperforming previous models.
- Notable strengths include tasks that involve visual reasoning, such as interpreting charts and graphs, and the ability to transcribe text from imperfect images.
New 'Artifacts' Feature
- The 'Artifacts' feature allows users to generate and interact with various types of AI-generated content, such as code snippets and website design.
- This preview feature marks Claude’s evolution from a conversational AI to a collaborative work environment.
|
Access |
|
Claude 3.5 Sonnet is available at no cost on claude.ai and the iOS app, with extended features available to Claude Pro and Team subscribers via Anthropic API, Amazon Bedrock, and Google Cloud’s Vertex AI. |
|
|
| TRY CLAUDE 3.5 |
|
|
|
 |
Start Selling to Enterprises with Just a Few Lines of Code |
|
Implement features like single sign-on in minutes instead of months.
-
WorkOS provides a complete User Management solution along with SSO, SCIM, and FGA.
-
Modular and easy-to-use APIs allow integrations to be completed in minutes instead of months.
-
Design and user experience are everything. From the quality of our documentation to how your users onboard, we remove all the unnecessary complexity for developers.
-
Best of all, User Management is free up to 1 million MAUs.
-
WorkOS powers some of the fastest growing AI startups like Perplexity, Copy.ai, and Cursor.
|
| GET STARTED |
|
partner with us |
|
|
|
TRENDING SIGNALS |
|
AGI |
|
|
|
⇧ 27,391 Likes |
|
|
|
Open Source |
|
|
|
⇧ 5311 Likes |
|
|
|
Agents |
|
|
|
⇧ 1844 Likes |
|
|
|
Numpy |
|
|
|
⇧ 2060 Likes |
|
|
|
AI Tools |
|
|
|
⇧ 190 Likes |
|
|
|
|
|
|
|
|
Learn AI/ML from first principles |
|
Join a class of experienced software engineers and level up your machine learning skills. After a few intensive weeks, you’ll be ready to solve problems with ML and ship ML-powered features.
This course is deeply technical and is only open to experienced developers. |
|
Apply Now ↗️ |
|
|
|
TOP REPOS |
|
Language Models |
|
|
|
☆ 926 Stars |
|
Tokencost helps calculate the USD cost of using major Large Language Model (LLMs) APIs by calculating the estimated cost of prompts and completions.
|
|
|
|
Diffusion |
|
|
|
☆ 39,454 Stars |
|
The most powerful and modular stable diffusion GUI, api and backend with a graph/nodes interface. This UI will let you design and execute advanced stable diffusion pipelines using a graph/nodes/flowchart based interface. |
|
|
|
Tutorials |
|
|
|
☆ 2491 Stars |
|
A powerful data & AI notebook templates catalog: prompts, plugins, models, workflow automation, analytics, code snippets - following the IMO framework to be searchable and reusable in any context. |
|
|
|
|
|
|
|
|
PYTORCH TIP |
Efficient Feature Engineering with Pandas apply and NumPy Vectorization |
|
Feature engineering is critical in machine learning, but it can be slow when processing large datasets. Many data scientists use pandas' apply method for row-wise operations, which can be inefficient. A faster alternative is leveraging NumPy's vectorization.
Instead of using pandas' apply method, use NumPy's vectorized operations to improve the speed of feature engineering tasks.
Why This Works
NumPy operations are implemented in C, which makes them much faster than the equivalent operations in pure Python. By using NumPy arrays and vectorized operations, you reduce the overhead associated with pandas' apply method, resulting in significant performance improvements.
Performance Comparison
To illustrate the performance gain, let's compare the execution time of both methods: |
import numpy as np
import pandas as pd
import time
# Define the complex function
def complex_function(x1, x2):
return np.log(x1 + 1) * np.sqrt(x2)
# Generate sample data
df = pd.DataFrame({
'feature1': np.random.rand(1000000),
'feature2': np.random.rand(1000000)
})
# Using pandas apply (slower)
start_time = time.time()
df['new_feature_apply'] = df.apply(lambda row: complex_function(
row['feature1'], row['feature2']), axis=1)
apply_duration = time.time() - start_time
# Using NumPy vectorization (faster)
start_time = time.time()
df['new_feature_vectorized'] = complex_function(df['feature1'].values,
df['feature2'].values)
vectorized_duration = time.time() - start_time
print(f"Apply duration: {apply_duration:.2f} seconds")
print(f"Vectorized duration: {vectorized_duration:.2f} seconds")
#Apply duration: 10.07 seconds
#Vectorized duration: 0.01 seconds
|
|
|
|
|