Pinned Loading

1. TimeSeriesForecasting TimeSeriesForecasting Public

   A production-grade benchmark framework designed to stress-test deep time-series models (DLinear, Linear_v2) against zero-parameter baselines. It simulates live telemetry loops by streaming sequence…

   Python

2. ChannelRegeneration ChannelRegeneration Public

   Python

3. DepthPruning-vLLM DepthPruning-vLLM Public

   Deployment-aware transformer compression pipeline using structured depth pruning, knowledge distillation, and vLLM serving. Benchmarks analyze latency, throughput, concurrency scaling, and KV-cache…

4. InferenceScaling-Qwen InferenceScaling-Qwen Public

   Investigating how inference-time compute influences reasoning performance in large language models through systematic scaling of self-consistency decoding.

   Python

5. Triton-FlashAttention Triton-FlashAttention Public

   GQA/MQA-extended FlashAttention-2 in Triton, with a KV-cache decode kernel — built on a public tutorial baseline, hardened for correctness (4 bugs found/fixed), and benchmarked against PyTorch SDPA…

   Python