Average Ratings 0 Ratings

Total
ease
features
design
support

No User Reviews. Be the first to provide a review:

Write a Review

Average Ratings 0 Ratings

Total
ease
features
design
support

No User Reviews. Be the first to provide a review:

Write a Review

Description

Recent breakthroughs in natural language processing, comprehension, and generation have been greatly influenced by the development of large language models. This research presents a system that employs Ascend 910 AI processors and the MindSpore framework to train a language model exceeding one trillion parameters, specifically 1.085 trillion, referred to as PanGu-{\Sigma}. This model enhances the groundwork established by PanGu-{\alpha} by converting the conventional dense Transformer model into a sparse format through a method known as Random Routed Experts (RRE). Utilizing a substantial dataset of 329 billion tokens, the model was effectively trained using a strategy called Expert Computation and Storage Separation (ECSS), which resulted in a remarkable 6.3-fold improvement in training throughput through the use of heterogeneous computing. Through various experiments, it was found that PanGu-{\Sigma} achieves a new benchmark in zero-shot learning across multiple downstream tasks in Chinese NLP, showcasing its potential in advancing the field. This advancement signifies a major leap forward in the capabilities of language models, illustrating the impact of innovative training techniques and architectural modifications.

Description

Qwen3.8-Flash-Next represents an open-weight multimodal Mixture-of-Experts architecture and serves as an initial glimpse into the design intended for Qwen4. This model strategically enhances attention mechanisms, residual pathways, embeddings, and optimization techniques to boost its capabilities, improve computational efficiency, expand model capacity, and ensure training stability. Its innovative hybrid architecture merges Gated DeltaNet, which adeptly compresses past information, with Qwen Sparse Attention, enabling the selection of significant context at a micro-block level to lessen both attention and indexing costs associated with lengthy sequences. The Gated Residual feature broadens the residual pathway into four streams, dynamically managing the flow of information across different layers. Additionally, the N-gram Embedding integrates large-scale local-pattern memory with minimal added computation per token, and it can be transferred to host memory for further efficiency. The model is structured around a 125B-parameter main network supplemented by 51B parameters dedicated to N-gram embeddings, activating only 6B parameters for each token processed. This sophisticated framework highlights the ongoing advancements in machine learning architectures, setting a promising stage for future developments.

API Access

Has API No 

API Access

Has API Yes 

Screenshots View All

No images available

Screenshots View All

Integrations

Alibaba Cloud No 
Cherry Studio No 
Cline No 
ClinePass No 
Happy Shrimp 1.0 No 
Hermes Agent No 
Model Context Protocol (MCP) No 
ModelScope No 
Novita AI No 
Odysseus No 
OfoxAI No 
Ollama No 
OpenClaw No 
PanGu Chat Yes 
Python No 
Qwen No 
Qwen Code No 
Qwen Studio No 
QwenCloud No 
QwenWork No 

Integrations

Alibaba Cloud Yes 
Cherry Studio Yes 
Cline Yes 
ClinePass Yes 
Happy Shrimp 1.0 Yes 
Hermes Agent Yes 
Model Context Protocol (MCP) Yes 
ModelScope Yes 
Novita AI Yes 
Odysseus Yes 
OfoxAI Yes 
Ollama Yes 
OpenClaw Yes 
PanGu Chat No 
Python Yes 
Qwen Yes 
Qwen Code Yes 
Qwen Studio Yes 
QwenCloud Yes 
QwenWork Yes 

Pricing Details

No price information available.
Free Trial No 
Free Version No 

Pricing Details

$2 per 1M (input)
Free Trial No 
Free Version No 

Deployment

Web-Based Yes 
On-Premises Yes 
iPhone App No 
iPad App No 
Android App No 
Windows No 
Mac No 
Linux No 
Chromebook No 

Deployment

Web-Based Yes 
On-Premises No 
iPhone App No 
iPad App No 
Android App No 
Windows No 
Mac No 
Linux No 
Chromebook No 

Customer Support

Business Hours No 
Live Rep (24/7) No 
Online Support No 

Customer Support

Business Hours No 
Live Rep (24/7) No 
Online Support Yes 

Types of Training

Training Docs Yes 
Webinars No 
Live Training (Online) No 
In Person No 

Types of Training

Training Docs Yes 
Webinars No 
Live Training (Online) No 
In Person No 

Vendor Details

Company Name

Huawei

Founded

1987

Country

China

Website

huawei.com

Vendor Details

Company Name

Alibaba

Founded

1999

Country

China

Website

qwen.ai/blog

Product Features

Alternatives

PanGu-α Reviews

PanGu-α

Huawei

Alternatives

LTM-1 Reviews

LTM-1

Magic AI
VideoPoet Reviews

VideoPoet

Google
Qwen3.5 Reviews

Qwen3.5

Alibaba
DeepSeek-V2 Reviews

DeepSeek-V2

DeepSeek