SILX AI
Popular repositories Loading
-
quasar-attention
quasar-attention PublicQuasar Attention Continuous-Time Memory Dynamics for Scaling Models Context to Millions of Tokens
-
QUASAR-SUBNET
QUASAR-SUBNET PublicQUASAR is a long-context foundation model and decentralized evaluation subnet built on Bittensor,
-
FlowAttention
FlowAttention PublicFlow Attention is a linear-time alternative to traditional Multi-Head Attention. Instead of relying on Query-Key-Value matrices and O(n²) pairwise interactions, it uses learned bidirectional flow d…
Repositories
- QUASAR-SUBNET Public
QUASAR is a long-context foundation model and decentralized evaluation subnet built on Bittensor,
SILX-LABS/QUASAR-SUBNET’s past year of commit activity - quasar-flash-linear-attention Public Forked from fla-org/flash-linear-attention
🚀 Efficient implementations of state-of-the-art linear attention models
SILX-LABS/quasar-flash-linear-attention’s past year of commit activity - Quasar-2.0 Public
Quasar 2.0 — Uses distillation to transfer knowledge from frozen teacher models (GLM, QWEN) to a decentralized Quasar model with MoE-based decentralized training.
SILX-LABS/Quasar-2.0’s past year of commit activity - quasar-attention Public
Quasar Attention Continuous-Time Memory Dynamics for Scaling Models Context to Millions of Tokens
SILX-LABS/quasar-attention’s past year of commit activity - SI-AGENT Public
SILX-LABS/SI-AGENT’s past year of commit activity - FlowAttention Public
Flow Attention is a linear-time alternative to traditional Multi-Head Attention. Instead of relying on Query-Key-Value matrices and O(n²) pairwise interactions, it uses learned bidirectional flow dynamics to propagate contextual information across sequences
SILX-LABS/FlowAttention’s past year of commit activity - TARS Public
SILX-LABS/TARS’s past year of commit activity - Quasar Public
Quasar-1 is a LLM architecture that uses a Token Temperature Mechanism to refine user inputs by distinguishing between 'hot' (critical) and 'cold' (less relevant) tokens. With Token Temperature-Based Attention Modulation and Prompt Refinement Loops, Quasar-1 enhances accuracy and task-specific performance for more adaptive, context-aware responses.
SILX-LABS/Quasar’s past year of commit activity
People
This organization has no public members. You must be a member to see who’s a part of this organization.
Top languages
Loading…
Most used topics
Loading…