Group 15: Reinforcement Learning for MASE Search Action

DeepWok / mase

Machine-Learning Accelerator System Exploration Tools

Other

117 stars 52 forks source link

Group 15: Reinforcement Learning for MASE Search Action #70

Closed ih8gin closed 6 months ago

ih8gin commented 6 months ago

Software: Added RLSearchSpaceMixedPrecisionPTQ Implemented SearchStrategyRL Implemented MixedPrecisionEnv Added RL_MixPrecisionQuantize.toml as configuration example Improved calculate_avg_bits_module_analysis_pass to take torch.nn.Conv2d into account

ih8gin commented 6 months ago

Format problems found in CI. Already fixed in latest commit.

ih8gin commented 6 months ago

Reformat again