057cc54b66
Track execution time of individual tensor operations (views, copies, reshapes etc) during LLM forward passes using CGo bindings to the native graph runtime. This helps identify performance bottlenecks in the computation graph and optimize memory operations that can significantly impact inference latency. |
||
|---|---|---|
| .. | ||
| cache.go | ||
| causal.go | ||
| causal_test.go | ||
| encoder.go | ||
| wrapper.go | ||