fMRI-foundation-model/src/prep_HCP_downstream.py

#!/usr/bin/env python
# coding: utf-8

# In[1]:


# Import packages and setup gpu configuration.
# This code block shouldnt need to be adjusted other than the model_name (if interactive)!
import os
import sys
import json
import yaml
import numpy as np
import math
import time
import datetime
import random
from tqdm import tqdm
import webdataset as wds
import matplotlib.pyplot as plt
import pandas as pd
import torch
import torch.nn as nn
from torchvision import transforms
import utils
from mae_utils import flat_models
from elbow.sinks import BufferedParquetWriter

## MODEL TO LOAD ##
if utils.is_interactive():
    model_name = "NSDflat_large_gsrFalse_"
else:
    model_name = sys.argv[1]
outdir = os.path.abspath(f'checkpoints/{model_name}')
print("outdir", outdir)

# Load previously saved config.yaml made during main training script
assert os.path.exists(f"{outdir}/config.yaml")
config = yaml.load(open(f"{outdir}/config.yaml", 'r'), Loader=yaml.FullLoader)
print(f"Loaded config.yaml from ckpt folder {outdir}")
# create global variables from the config
print("\n__CONFIG__")
for attribute_name in config.keys():
    print(f"{attribute_name} = {config[attribute_name]}")
    globals()[attribute_name] = config[f'{attribute_name}']
print("\n")

if utils.is_interactive():
    # Following allows you to change functions in other files and 
    # have this notebook automatically update with your revisions
    get_ipython().run_line_magic('load_ext', 'autoreload')
    get_ipython().run_line_magic('autoreload', '2')

device = torch.device('cuda')

print("PID of this process =",os.getpid())

# seed all random functions
utils.seed_everything(seed)


# In[2]:


os.environ['HCP_FLAT_ROOT'] = hcp_flat_path


# In[3]:


if os.getenv('global_pool') == "False":
    global_pool = False
else:
    global_pool = True
print(f"global_pool = {global_pool}")

try:
    gsr
except:
    gsr = True
    print("set gsr to True")
print(f"gsr = {gsr}")


# # hcp_flat

# In[4]:


from mae_utils.flat import load_hcp_flat_mask
from mae_utils.flat import create_hcp_flat
from mae_utils.flat import batch_unmask
import mae_utils.visualize as vis

flat_mask = load_hcp_flat_mask(hcp_flat_path)

model = flat_models.mae_vit_large_fmri(
    patch_size=patch_size,
    decoder_embed_dim=decoder_embed_dim,
    t_patch_size=t_patch_size,
    pred_t_dim=pred_t_dim,
    decoder_depth=4,
    cls_embed=cls_embed,
    norm_pix_loss=norm_pix_loss,
    no_qkv_bias=no_qkv_bias,
    sep_pos_embed=sep_pos_embed,
    trunc_init=trunc_init,
    pct_masks_to_decode=pct_masks_to_decode,
    img_mask=flat_mask,
)


# # Load checkpoint

# In[5]:


checkpoint_files = [f for f in os.listdir(outdir) if f.endswith('.pth')]

if utils.is_interactive():
    latest_checkpoint = "epoch99.pth"
else:
    latest_checkpoint = sys.argv[2] 
print(f"latest_checkpoint: {latest_checkpoint}")

# Load the checkpoint
checkpoint_path = os.path.join(outdir, latest_checkpoint)

state = torch.load(checkpoint_path)
model.load_state_dict(state["model_state_dict"], strict=False)
model.to(device)
model.eval()

print(f"\nLoaded checkpoint {latest_checkpoint} from {outdir}\n")


# ## Create dataset and data loaders

# In[6]:


from torch.utils.data import default_collate
batch_size = 1
print(f"changed batch_size to {batch_size}")

## Test ##
datasets_to_include = "HCP"
assert "HCP" in datasets_to_include
test_dataset = create_hcp_flat(root=hcp_flat_path, 
                clip_mode="event", frames=num_frames, shuffle=False, gsr=gsr, sub_list = 'test')
test_dl = wds.WebLoader(
    test_dataset.batched(batch_size, partial=False, collation_fn=default_collate),
    batch_size=None,
    shuffle=False,
    num_workers=num_workers,
    pin_memory=True,
)

## Train ##
assert "HCP" in datasets_to_include
train_dataset = create_hcp_flat(root=hcp_flat_path, 
                clip_mode="event", frames=num_frames, shuffle=False, gsr=gsr, sub_list = 'train')
train_dl = wds.WebLoader(
    train_dataset.batched(batch_size, partial=False, collation_fn=default_collate),
    batch_size=None,
    shuffle=False,
    num_workers=num_workers,
    pin_memory=True,
)


# # Start extraction

# In[7]:


cnt = 9999 # need to change this


# In[8]:


@torch.no_grad()
def extract_features(dl, global_pool=True):
    for samples in tqdm(dl,total=cnt): 
        samples_meta = samples['meta']
        features = model(samples['image'].to(device),global_pool=global_pool, forward_features = True)
        features = features.flatten(1)
        features = features.cpu().numpy()
        meta_dict = {}
        for key, value in samples_meta.items():
            if type(value) == torch.Tensor:
                value = value.cpu().numpy()
            meta_dict[key] = value
        for feat, meta in zip(features, samples_meta):
            yield {"feature": feat, **meta_dict}


# In[9]:


out_folder = f'{outdir}_gp{global_pool}/{latest_checkpoint[:-4]}/HCP'
print(out_folder)
os.makedirs(out_folder,exist_ok=True)


# In[10]:


# Ensure the output Parquet directory exists
outdir_parquet = os.path.join(f'{outdir}_gp{global_pool}/{latest_checkpoint[:-4]}', 'HCP')
os.makedirs(outdir_parquet, exist_ok=True)  # <-- Add this line

utils.seed_everything(seed)

print("Start extract")
start_time = time.time()

with BufferedParquetWriter(f"{outdir_parquet}/test.parquet", blocking=True) as writer:
    for sample in extract_features(test_dl, global_pool):
        writer.write(sample)

with BufferedParquetWriter(f"{outdir_parquet}/train.parquet", blocking=True) as writer:
    for sample in extract_features(train_dl, global_pool):
        writer.write(sample)

total_time = time.time() - start_time
total_time_str = str(datetime.timedelta(seconds=int(total_time)))
print("Extract time {}".format(total_time_str))
print(torch.cuda.memory_allocated())


# In[ ]: