you guys did incredible work; the performance to cost ratio is so good!
HARDWARE
-
Scaling LLM Training: Double Digit Batch Size for 70B Models
By
–
double digit batch size, eventual goal is one rack for 70B.
-
SambaNova Cloud Launches Fastest AI Inference Platform Free
By
–
The fastest API for #devs is here — for FREE. This week, we launched SambaNova Cloud, the only platform that offers #AI inference of over 570 t/s on @AIatMeta
’s Llama 3.1 70B & 100 t/s for Llama 3.1 405B. Read more https://
sambanova.ai/blog/fastest-i
nference-best-models
… -
Oracle Announces Zettascale OCI Supercluster for AI Training
By
–
At #OCW24, @Oracle announced the first zettascale OCI Supercluster accelerated by the #NVIDIABlackwell platform, allowing customers to train and deploy next-generation #AI models at scale. Learn more: https://
nvda.ws/4ehTXAq -
Cerebras Inference Speed Update: Llama 3.1 Performance Benchmarks
By
–
Cerebras Inference perf update:
Llama3.1-8B: 1,8001,927 tokens/s
Llama3.1-70B: 450481 tokens/s
Stillfor the most popular open model in the world. https://
inference.cerebras.ai -
Coding Inside Vision Pro: New Development Possibilities
By
–
Or you could try coding inside Vision Pro
-
Flux LoRA Training Space Updated with L40S GPU for Speed
By
–
Flux LoRA Training Space has now been updated to use the L40S GPU. Now its even faster and cheaper to train your own Flux LoRA Training Space: https://
huggingface.co/spaces/autotra
in-projects/train-flux-lora-ease
… -
AutoTrain Bigger Models Faster with New L40S GPUs on Hugging Face
By
–
With the new L40S GPUs available in Hugging Face Spaces, you can now AutoTrain bigger models, faster
-
SambaNova Cloud Launch: Optimized AI Hardware Solution
By
–
We've loved the conversations around our newly launched SambaNova Cloud. In his article, @capacitymedia
's @benwodecki gives his insight: “The SambaNova Cloud is similar to services from rivals… however, the hardware is optimized to a point where it can run on a single -
imAIgine SDK Revolutionizes Neural Network Deployment to Hardware
By
–
Imagination reality. Our imAIgine® SDK revolutionizes the deployment of neural networks to HW, cutting process from weeks to minutes. Experience the ease of deploying AI. Learn about our EA release supporting speedAI® devices. https://
untether.ai/imaigine-sdk-e
arly-access-program/
… https://
youtu.be/qIqL405vXok