|
C-Transformer
Cache-Optimized Transformers in C (x86)
|
Public Attributes | |
| size_t | layer_start_canary_offset |
| size_t | ln1_weight_offset |
| size_t | ln1_bias_offset |
| size_t | ln1_mean_offset |
| size_t | ln1_rstd_offset |
| size_t | layer_input_offset |
| size_t | ln1_output_offset |
| size_t | q_weight_offset |
| size_t | q_bias_offset |
| size_t | q_output_offset |
| size_t | k_weight_offset |
| size_t | k_bias_offset |
| size_t | k_output_offset |
| size_t | v_weight_offset |
| size_t | v_bias_offset |
| size_t | v_output_offset |
| size_t | attention_scores_offset |
| size_t | proj_weight_offset |
| size_t | proj_bias_offset |
| size_t | attention_output_offset |
| size_t | residual1_output_offset |
| size_t | ln2_weight_offset |
| size_t | ln2_bias_offset |
| size_t | ln2_mean_offset |
| size_t | ln2_rstd_offset |
| size_t | ln2_output_offset |
| size_t | fc1_weight_offset |
| size_t | fc1_bias_offset |
| size_t | fc1_output_offset |
| size_t | fc2_weight_offset |
| size_t | fc2_bias_offset |
| size_t | mlp_output_offset |
| size_t | residual2_output_offset |
| size_t | layer_end_canary_offset |