| 1028 | } |
| 1029 | |
| 1030 | struct train_params_common get_default_train_params_common() { |
| 1031 | struct train_params_common params; |
| 1032 | params.fn_train_data = "shakespeare.txt"; |
| 1033 | params.fn_checkpoint_in = "checkpoint.gguf"; |
| 1034 | params.fn_checkpoint_out = "checkpoint-ITERATION.gguf"; |
| 1035 | params.pattern_fn_it = "ITERATION"; |
| 1036 | params.fn_latest = "LATEST"; |
| 1037 | |
| 1038 | params.print_usage = false; |
| 1039 | |
| 1040 | params.save_every = 10; |
| 1041 | |
| 1042 | params.seed = -1; |
| 1043 | |
| 1044 | params.n_ctx = 128; |
| 1045 | params.n_threads = 6; |
| 1046 | params.n_batch = 8; |
| 1047 | params.n_gradient_accumulation = 1; |
| 1048 | params.n_epochs = -1; |
| 1049 | params.n_gpu_layers = 0; |
| 1050 | |
| 1051 | params.custom_n_ctx = false; |
| 1052 | |
| 1053 | params.use_flash = true; |
| 1054 | params.use_checkpointing = true; |
| 1055 | |
| 1056 | params.sample_start = ""; |
| 1057 | params.include_sample_start = false; |
| 1058 | params.escape = false; |
| 1059 | params.overlapping_samples = false; |
| 1060 | params.fill_with_next_samples = false; |
| 1061 | params.separate_with_eos = false; |
| 1062 | params.separate_with_bos = true; |
| 1063 | params.sample_random_offsets = false; |
| 1064 | params.force_reshuffle = false; |
| 1065 | |
| 1066 | params.opt_past = 0; |
| 1067 | params.opt_delta = 1e-5f; |
| 1068 | params.opt_max_no_improvement = 0; |
| 1069 | |
| 1070 | params.warmup = 100; |
| 1071 | params.cos_decay_steps = 1000; |
| 1072 | params.cos_decay_restart = 1.1f; |
| 1073 | params.cos_decay_min = 0.1f; |
| 1074 | params.enable_restart = false; |
| 1075 | |
| 1076 | params.adam_n_iter = 256; |
| 1077 | params.adam_alpha = 1e-3f; |
| 1078 | params.adam_min_alpha = 0; |
| 1079 | params.adam_decay = 1e-1f; |
| 1080 | params.adam_decay_min_ndim = 2; |
| 1081 | params.adam_beta1 = 0.9f; |
| 1082 | params.adam_beta2 = 0.999f; |
| 1083 | params.adam_gclip = 1.0f; |
| 1084 | params.adam_eps_f = 0.0f; |
| 1085 | |
| 1086 | return params; |
| 1087 | } |
no outgoing calls
no test coverage detected