getting started
starting main
Executing MEMIT with parameters MEMITHyperParams(layers=[4, 5, 6, 7, 8], layer_selection='all', fact_token='subject_last', v_num_grad_steps=25, v_lr=0.5, v_loss_layer=31, v_weight_decay=0.001, clamp_norm_factor=4, kl_factor=0.0625, mom2_adjustment=True, mom2_update_weight=15000, rewrite_module_tmp='model.layers.{}.mlp.down_proj', layer_module_tmp='model.layers.{}', mlp_module_tmp='model.layers.{}.mlp', attn_module_tmp='model.layers.{}.self_attn', ln_f_module='model.norm', lm_head_module='lm_head', stats_dir='./data/stats', mom2_dataset='wikipedia', mom2_n_samples=100000, mom2_dtype='float32')
Instantiating model
Loading dataset, attribute snippets, tf-idf data
Will load cache from kvs/meta-llama_Llama-2-7b-hf_MEMIT/P27_P101_layer_{}_clamp_{}_case_{}.npz
working on results/llama/MEMIT/run_000/{}_edits-case_{}.json
about to start 23061
started 23061
Execution took 99797.50721216202
about to evaluate
getting started
starting main
Executing MEMIT with parameters MEMITHyperParams(layers=[4, 5, 6, 7, 8], layer_selection='all', fact_token='subject_last', v_num_grad_steps=25, v_lr=0.5, v_loss_layer=31, v_weight_decay=0.001, clamp_norm_factor=4, kl_factor=0.0625, mom2_adjustment=True, mom2_update_weight=15000, rewrite_module_tmp='model.layers.{}.mlp.down_proj', layer_module_tmp='model.layers.{}', mlp_module_tmp='model.layers.{}.mlp', attn_module_tmp='model.layers.{}.self_attn', ln_f_module='model.norm', lm_head_module='lm_head', stats_dir='./data/stats', mom2_dataset='wikipedia', mom2_n_samples=100000, mom2_dtype='float32')
Instantiating model
Loading dataset, attribute snippets, tf-idf data
Will load cache from kvs/meta-llama_Llama-2-7b-hf_MEMIT/P21_P101_layer_{}_clamp_{}_case_{}.npz
working on results/llama/MEMIT/run_001/{}_edits-case_{}.json
about to start 21920
started 21920
Execution took 253.0333116054535
about to evaluate
getting started
starting main
Executing MEMIT with parameters MEMITHyperParams(layers=[4, 5, 6, 7, 8], layer_selection='all', fact_token='subject_last', v_num_grad_steps=25, v_lr=0.5, v_loss_layer=31, v_weight_decay=0.001, clamp_norm_factor=4, kl_factor=0.0625, mom2_adjustment=True, mom2_update_weight=15000, rewrite_module_tmp='model.layers.{}.mlp.down_proj', layer_module_tmp='model.layers.{}', mlp_module_tmp='model.layers.{}.mlp', attn_module_tmp='model.layers.{}.self_attn', ln_f_module='model.norm', lm_head_module='lm_head', stats_dir='./data/stats', mom2_dataset='wikipedia', mom2_n_samples=100000, mom2_dtype='float32')
Instantiating model
Loading dataset, attribute snippets, tf-idf data
Will load cache from kvs/meta-llama_Llama-2-7b-hf_MEMIT/P19_P101_layer_{}_clamp_{}_case_{}.npz
working on results/llama/MEMIT/run_001/{}_edits-case_{}.json
about to start 22775
started 22775
Execution took 239.85058569908142
about to evaluate
getting started
starting main
Executing MEMIT with parameters MEMITHyperParams(layers=[4, 5, 6, 7, 8], layer_selection='all', fact_token='subject_last', v_num_grad_steps=25, v_lr=0.5, v_loss_layer=31, v_weight_decay=0.001, clamp_norm_factor=4, kl_factor=0.0625, mom2_adjustment=True, mom2_update_weight=15000, rewrite_module_tmp='model.layers.{}.mlp.down_proj', layer_module_tmp='model.layers.{}', mlp_module_tmp='model.layers.{}.mlp', attn_module_tmp='model.layers.{}.self_attn', ln_f_module='model.norm', lm_head_module='lm_head', stats_dir='./data/stats', mom2_dataset='wikipedia', mom2_n_samples=100000, mom2_dtype='float32')
Instantiating model
Loading dataset, attribute snippets, tf-idf data
Will load cache from kvs/meta-llama_Llama-2-7b-hf_MEMIT/P103_layer_{}_clamp_{}_case_{}.npz
working on results/llama/MEMIT/run_001/{}_edits-case_{}.json
about to start 0
started 0
Execution took 7660.189071655273
about to evaluate
getting started
starting main
Executing MEMIT with parameters MEMITHyperParams(layers=[4, 5, 6, 7, 8], layer_selection='all', fact_token='subject_last', v_num_grad_steps=25, v_lr=0.5, v_loss_layer=31, v_weight_decay=0.001, clamp_norm_factor=4, kl_factor=0.0625, mom2_adjustment=True, mom2_update_weight=15000, rewrite_module_tmp='model.layers.{}.mlp.down_proj', layer_module_tmp='model.layers.{}', mlp_module_tmp='model.layers.{}.mlp', attn_module_tmp='model.layers.{}.self_attn', ln_f_module='model.norm', lm_head_module='lm_head', stats_dir='./data/stats', mom2_dataset='wikipedia', mom2_n_samples=100000, mom2_dtype='float32')
Instantiating model
Loading dataset, attribute snippets, tf-idf data
Will load cache from kvs/meta-llama_Llama-2-7b-hf_MEMIT/P101_layer_{}_clamp_{}_case_{}.npz
working on results/llama/MEMIT/run_001/{}_edits-case_{}.json
about to start 73
started 73
Execution took 3284.5443892478943
about to evaluate
getting started
starting main
Instantiating model
Loading dataset, attribute snippets, tf-idf data
Will load cache from kvs/meta-llama_Llama-2-7b-hf_NONE/P101_P21_layer_{}_clamp_{}_case_{}.npz
about to evaluate
getting started
starting main
Instantiating model
Loading dataset, attribute snippets, tf-idf data
Will load cache from kvs/meta-llama_Llama-2-7b-hf_NONE/P27_P19_layer_{}_clamp_{}_case_{}.npz
about to evaluate
getting started
starting main
Instantiating model
Loading dataset, attribute snippets, tf-idf data
Will load cache from kvs/meta-llama_Llama-2-7b-hf_NONE/P101_P21_layer_{}_clamp_{}_case_{}.npz
getting started
starting main
Instantiating model
Loading dataset, attribute snippets, tf-idf data
Will load cache from kvs/meta-llama_Llama-2-7b-hf_NONE/P27_P19_layer_{}_clamp_{}_case_{}.npz
about to evaluate
getting started
starting main
Instantiating model
Loading dataset, attribute snippets, tf-idf data
Will load cache from kvs/meta-llama_Llama-2-7b-hf_NONE/P21_P101_layer_{}_clamp_{}_case_{}.npz
about to evaluate
getting started
starting main
Instantiating model
Loading dataset, attribute snippets, tf-idf data
Will load cache from kvs/meta-llama_Llama-2-7b-hf_NONE/P27_P101_layer_{}_clamp_{}_case_{}.npz
about to evaluate
getting started
starting main
Instantiating model
Loading dataset, attribute snippets, tf-idf data
Will load cache from kvs/meta-llama_Llama-2-7b-hf_NONE/P27_P21_layer_{}_clamp_{}_case_{}.npz
about to evaluate
getting started
starting main
Instantiating model
Loading dataset, attribute snippets, tf-idf data
Will load cache from kvs/meta-llama_Llama-2-7b-hf_NONE/P27_P19_layer_{}_clamp_{}_case_{}.npz
getting started
starting main
Instantiating model
Loading dataset, attribute snippets, tf-idf data
Will load cache from kvs/meta-llama_Llama-2-7b-hf_NONE/P101_P27_layer_{}_clamp_{}_case_{}.npz
about to evaluate
getting started
starting main
Instantiating model
Loading dataset, attribute snippets, tf-idf data
Will load cache from kvs/meta-llama_Llama-2-7b-hf_NONE/P19_P21_layer_{}_clamp_{}_case_{}.npz
about to evaluate
getting started
starting main
Instantiating model
Loading dataset, attribute snippets, tf-idf data
Will load cache from kvs/meta-llama_Llama-2-7b-hf_NONE/P19_P101_layer_{}_clamp_{}_case_{}.npz
about to evaluate
