-
Notifications
You must be signed in to change notification settings - Fork 6
/
Copy pathMakefile
138 lines (117 loc) · 8.06 KB
/
Makefile
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
.PHONY: results, results_no_free, learning_curves, eval_results
data/bayes_nets/nets_n-$(N_NETS)_nodes-$(N_NODES)_edges-$(N_EDGES).pkl: code/make_train_data/generate_bayes_nets.py
python code/make_train_data/generate_bayes_nets.py \
--n_nets $(N_NETS) \
--n_nodes $(N_NODES) \
--n_edges $(N_EDGES)
data/evaluation/true-probs/true-probabilities-net-$(NET_ID).csv: code/evaluate/true_conditional_probs.py data/bayes_nets/nets_n-$(N_NETS)_nodes-$(N_NODES)_edges-$(N_EDGES).pkl
python code/evaluate/true_conditional_probs.py \
--net_idx $(NET_ID) \
--bayes-net-file data/bayes_nets/nets_n-$(N_NETS)_nodes-$(N_NODES)_edges-$(N_EDGES).pkl
data/training-data/selected-pairs/selected-pairs-net-$(NET_ID).csv: data/evaluation/true-probs/true-probabilities-net-$(NET_ID).csv code/make_train_data/select_pairs_to_hold_out.py
python3 code/make_train_data/select_pairs_to_hold_out.py \
--net_idx $(NET_ID) \
--n_pairs $(NUM_PAIRS)
data/scaffolds/scaffolds-net-$(NET_ID).csv: code/scaffold/generate_scaffolds.py data/evaluation/true-probs/true-probabilities-net-$(NET_ID).csv data/bayes_nets/nets_n-$(N_NETS)_nodes-$(N_NODES)_edges-$(N_EDGES).pkl
python code/scaffold/generate_scaffolds.py \
--net-idx $(NET_ID) \
--num-scaffolds $(NUM_PAIRS) \
--bayes-net-file data/bayes_nets/nets_n-$(N_NETS)_nodes-$(N_NODES)_edges-$(N_EDGES).pkl
data/scaffolds/negative-scaffolds-net-$(NET_ID).csv: code/scaffold/generate_scaffolds.py code/scaffold/generate_negative_scaffolds.py data/evaluation/true-probs/true-probabilities-net-$(NET_ID).csv data/bayes_nets/nets_n-$(N_NETS)_nodes-$(N_NODES)_edges-$(N_EDGES).pkl
python code/scaffold/generate_negative_scaffolds.py \
--net-idx $(NET_ID) \
--num-scaffolds $(NUM_PAIRS) \
--bayes-net-file data/bayes_nets/nets_n-$(N_NETS)_nodes-$(N_NODES)_edges-$(N_EDGES).pkl
data/training-data/samples/train_samples_$(SAMPLE_FORMAT_STR)_net_$(NET_ID).csv: data/bayes_nets/nets_n-$(N_NETS)_nodes-$(N_NODES)_edges-$(N_EDGES).pkl # data/training-data/selected-pairs/selected-pairs-net-$(NET_ID).csv
cd code/make_train_data && python generate_training_set.py \
-n $(N_TRAIN) \
--sample-format $(SAMPLE_FORMAT) \
--sample-format-str $(SAMPLE_FORMAT_STR) \
--net-id $(NET_ID) \
--exp-p $(EXP_P) \
--zipf-k $(ZIPF_K) \
--bayes-net-file data/bayes_nets/nets_n-$(N_NETS)_nodes-$(N_NODES)_edges-$(N_EDGES).pkl
$(MODEL_ROOT_FOLDER)/$(MODEL_NAME)/pytorch_model.bin: data/training-data/samples/train_samples_$(SAMPLE_FORMAT_STR)_net_$(NET_ID).csv
python code/finetune/run_clm.py \
--model_name_or_path $(BASE_MODEL_PATH) \
--train_file data/training-data/samples/train_samples_$(SAMPLE_FORMAT_STR)_net_$(NET_ID).csv \
--per_device_train_batch_size 3 \
--per_device_eval_batch_size 3 \
--save_total_limit $(TOTAL_CHECKPOINTS) \
--save_steps $(CHECKPOINT_INTERVAL) \
--do_train \
--num_train_epochs $(N_EPOCHS) \
--max_steps $(N_TRAIN_STEPS) \
--output_dir $(MODEL_ROOT_FOLDER)/$(MODEL_NAME)
data/samples/$(MODEL_NAME)_raw.csv: $(MODEL_ROOT_FOLDER)/$(MODEL_NAME)/pytorch_model.bin code/generate/generate_from_finetuned_model.py
CUDA_VISIBLE_DEVICES=0 python code/generate/generate_from_finetuned_model.py \
--n_generations 100000 \
--model_folder $(MODEL_ROOT_FOLDER)/$(MODEL_NAME) \
--prefix_style "random_conditions_and_targets" \
--n_vars $(N_NODES)
data/evaluation/base-model-$(BASE_MODEL_NAME)/fixed-gen-probabilities-$(MODEL_NAME).csv: $(MODEL_ROOT_FOLDER)/$(MODEL_NAME)/pytorch_model.bin code/evaluate/fixed_generation_probabilities.py
python code/evaluate/fixed_generation_probabilities.py \
--model_folder $(MODEL_ROOT_FOLDER)/$(MODEL_NAME) \
--base_model_name $(BASE_MODEL_NAME) \
--net_idx $(NET_ID) \
--device "cuda:0" \
--bayes-net-file data/bayes_nets/nets_n-$(N_NETS)_nodes-$(N_NODES)_edges-$(N_EDGES).pkl
data/evaluation/base-model-$(BASE_MODEL_NAME)/free-gen-probabilities-$(MODEL_NAME)-$(NUM_SAMPLES)samples.csv: $(MODEL_ROOT_FOLDER)/$(MODEL_NAME)/pytorch_model.bin code/evaluate/free_generation_probabilities.py
python code/evaluate/free_generation_probabilities.py \
--model_folder $(MODEL_ROOT_FOLDER)/$(MODEL_NAME) \
--base_model_name $(BASE_MODEL_NAME) \
--net_idx $(NET_ID) \
--device "cuda:0" \
--num_samples $(NUM_SAMPLES) \
--bayes-net-file data/bayes_nets/nets_n-$(N_NETS)_nodes-$(N_NODES)_edges-$(N_EDGES).pkl
data/evaluation/base-model-$(BASE_MODEL_NAME)/scaffolded-gen-probabilities-$(MODEL_NAME)-$(NUM_SAMPLES)samples.csv: $(MODEL_ROOT_FOLDER)/$(MODEL_NAME)/pytorch_model.bin data/scaffolds/scaffolds-net-$(NET_ID).csv code/evaluate/scaffolded_generation_probabilities.py data/bayes_nets/nets_n-$(N_NETS)_nodes-$(N_NODES)_edges-$(N_EDGES).pkl
python code/evaluate/scaffolded_generation_probabilities.py \
--model_folder $(MODEL_ROOT_FOLDER)/$(MODEL_NAME) \
--base_model_name $(BASE_MODEL_NAME) \
--scaffold_file data/scaffolds/scaffolds-$(MODEL_NAME).json \
--net_idx $(NET_ID) \
--device "cuda:0" \
--num_samples $(NUM_SAMPLES) \
--bayes-net-file data/bayes_nets/nets_n-$(N_NETS)_nodes-$(N_NODES)_edges-$(N_EDGES).pkl
data/evaluation/base-model-$(BASE_MODEL_NAME)/negative-scaffolded-gen-probabilities-$(MODEL_NAME)-$(NUM_SAMPLES)samples.csv: $(MODEL_ROOT_FOLDER)/$(MODEL_NAME)/pytorch_model.bin data/scaffolds/negative-scaffolds-net-$(NET_ID).csv code/evaluate/scaffolded_generation_probabilities.py data/bayes_nets/nets_n-$(N_NETS)_nodes-$(N_NODES)_edges-$(N_EDGES).pkl
python code/evaluate/scaffolded_generation_probabilities.py \
--model_folder $(MODEL_ROOT_FOLDER)/$(MODEL_NAME) \
--base_model_name $(BASE_MODEL_NAME) \
--scaffold_file data/scaffolds/scaffolds-$(MODEL_NAME).json \
--net_idx $(NET_ID) \
--device "cuda:0" \
--num_samples $(NUM_SAMPLES) \
--bayes-net-file data/bayes_nets/nets_n-$(N_NETS)_nodes-$(N_NODES)_edges-$(N_EDGES).pkl \
--negative
data/evaluation/base-model-$(BASE_MODEL_NAME)/learning-curves/learning-curves-$(MODEL_NAME).csv: code/evaluate/make_learning_curves.py code/evaluate/free_generation_probabilities.py code/evaluate/fixed_generation_probabilities.py
python code/evaluate/make_learning_curves.py \
--model_folder $(MODEL_ROOT_FOLDER)/$(MODEL_NAME) \
--net_idx $(NET_ID) \
--device "cuda:0" \
--num_samples $(NUM_SAMPLES) \
--bayes-net-file data/bayes_nets/nets_n-$(N_NETS)_nodes-$(N_NODES)_edges-$(N_EDGES).pkl
data/evaluation/base-model-$(BASE_MODEL_NAME)/losses/losses-$(MODEL_NAME).csv: code/evaluate/compile_training_losses.py $(MODEL_ROOT_FOLDER)/$(MODEL_NAME)/pytorch_model.bin
python code/evaluate/compile_training_losses.py \
--model_folder $(MODEL_ROOT_FOLDER)/$(MODEL_NAME) \
--base_arch $(BASE_MODEL_NAME)
data/evaluation/base-model-$(BASE_MODEL_NAME)/eval-results/eval-results-$(MODEL_NAME).csv: code/evaluate/compile_eval_results.py
python code/evaluate/compile_eval_results.py \
--model_folder $(MODEL_ROOT_FOLDER)/$(MODEL_NAME) \
--base_arch $(BASE_MODEL_NAME)
$(MODEL_ROOT_FOLDER)/$(MODEL_NAME)/eval_results.json: $(MODEL_ROOT_FOLDER)/$(MODEL_NAME)/pytorch_model.bin
python code/finetune/compute_perplexities.py \
--model_folder $(MODEL_ROOT_FOLDER)/$(MODEL_NAME) \
--base_arch $(BASE_MODEL_NAME) \
--include_checkpoints
results: data/evaluation/true-probs/true-probabilities-net-$(NET_ID).csv \
data/evaluation/base-model-$(BASE_MODEL_NAME)/fixed-gen-probabilities-$(MODEL_NAME).csv \
data/evaluation/base-model-$(BASE_MODEL_NAME)/free-gen-probabilities-$(MODEL_NAME)-$(NUM_SAMPLES)samples.csv \
data/evaluation/base-model-$(BASE_MODEL_NAME)/scaffolded-gen-probabilities-$(MODEL_NAME)-$(NUM_SAMPLES)samples.csv \
data/evaluation/base-model-$(BASE_MODEL_NAME)/negative-scaffolded-gen-probabilities-$(MODEL_NAME)-$(NUM_SAMPLES)samples.csv \
$(MODEL_ROOT_FOLDER)/$(MODEL_NAME)/pytorch_model.bin \
data/evaluation/base-model-$(BASE_MODEL_NAME)/losses/losses-$(MODEL_NAME).csv
learning_curves: $(MODEL_ROOT_FOLDER)/$(MODEL_NAME)/pytorch_model.bin \
data/evaluation/base-model-$(BASE_MODEL_NAME)/learning-curves/learning-curves-$(MODEL_NAME).csv \
data/evaluation/base-model-$(BASE_MODEL_NAME)/losses/losses-$(MODEL_NAME).csv
eval_results: $(MODEL_ROOT_FOLDER)/$(MODEL_NAME)/eval_results.json \
data/evaluation/base-model-$(BASE_MODEL_NAME)/eval-results/eval-results-$(MODEL_NAME).csv