Two-stage working example.

Author: zeisler

dvc.lock

@@ -4,18 +4,12 @@ stages:
     cmd: python src/get_data.py --output_folder outs/data
     deps:
     - path: src/get_data.py
-      md5: 0b9160f7475a6aaa50acb793b7527f68
-      size: 991
+      md5: c7452de42878b5e9e672d786d48e7c53
+      size: 998
     params:
       params.yaml:
         data:
-          repo: iterative/dvc
-          labels:
-          - data-sync
-          - experiments
-          - plots
-          since: 2021/01/01
-          until: 2022/05/01
+          data_date: 2022/08/08
           output_folder: outs/data
           metrics_file: outs/data_metrics.json
     outs:
@@ -118,3 +112,32 @@ stages:
     - path: outs/eval/plots/confusion_matrix.json
       md5: 7ad7650d9b00d4ae671f04ba20889b79
       size: 1274
+  fit:
+    cmd: python src/fit.py outs/data outs/fit outs/fit_metrics
+    deps:
+    - path: outs/data
+      md5: 0223fa49fdf31fd4bb2f99cf8322f401.dir
+      size: 136
+      nfiles: 1
+    - path: src/fit.py
+      md5: ef5f1cb19e0a3d286c43877e6b459620
+      size: 947
+    params:
+      params.yaml:
+        fit:
+          output_folder: outs/fit
+          metrics_folder: outs/fit_metrics
+          metrics_file: outs/fit_metrics.json
+          epochs: 8
+    outs:
+    - path: outs/fit
+      md5: d751713988987e9331980363e24189ce.dir
+      size: 0
+      nfiles: 0
+    - path: outs/fit_metrics
+      md5: d751713988987e9331980363e24189ce.dir
+      size: 0
+      nfiles: 0
+    - path: outs/fit_metrics.json
+      md5: eba383f3de32d0a53c5962f4119e6dcf
+      size: 23

dvc.yaml

@@ -18,58 +18,22 @@ stages:
       - ${data.metrics_file}:
           cache: false
 
-  split-data:
+  fit:
     cmd: 
-      python src/split_data.py 
-      ${data.output_folder} 
-      ${split.output_folder} 
-      ${split.test_size}
-      ${split.metrics_file} 
+      python src/fit.py
+      ${data.output_folder}
+      ${fit.output_folder}
+      ${fit.metrics_folder}
     deps:
     - ${data.output_folder}
-    - src/split_data.py
-    outs:
-    - ${split.output_folder}
-    metrics:
-    - ${split.metrics_file}:
-        cache: false
-
-  train:
-    cmd: 
-      python src/train.py
-      ${split.output_folder}
-      ${train.output_folder}
-    deps:
-    - ${split.output_folder}
-    - src/train.py
+    - src/fit.py
     params:
-      - data.labels
-      - train
+      - fit
     outs:
-    - ${train.output_folder}
+    - ${fit.output_folder}
     metrics:
-      - ${train.metrics_folder}.json:
+      - ${fit.metrics_folder}.json:
           cache: false
     plots:
-      - ${train.metrics_folder}:
+      - ${fit.metrics_folder}:
           cache: false
-
-  eval:
-    cmd: 
-      python src/eval.py
-      ${split.output_folder}/val.json
-      ${train.output_folder}
-      ${eval.output_folder}
-    params:
-      - data.labels
-    deps:
-    - ${split.output_folder}/val.json
-    - ${train.output_folder}
-    - src/eval.py
-    - src/inference.py
-    plots:
-    - ${eval.output_folder}/plots/confusion_matrix.json:
-        cache: false
-        template: confusion
-        x: actual
-        y: predicted

outs/.gitignore

@@ -1,3 +1,4 @@
 /data
 /split
 /train
+/fit

outs/fit_metrics.json

@@ -0,0 +1,3 @@
+{
+    "num_trades": 2
+}

params.yaml

@@ -1,26 +1,13 @@
 data:
-  repo: iterative/dvc
-
-  labels:
-  - data-sync
-  - experiments
-  - plots
-  since: 2021/01/01
-  until: 2022/05/01
+  data_date: 2022/08/08
 
   output_folder: outs/data
   metrics_file: outs/data_metrics.json
 
-split:
-  output_folder: outs/split
-  metrics_file: outs/split_metrics.json
-  test_size: 0.1
+fit:
+  output_folder: outs/fit
+  metrics_folder: outs/fit_metrics
+  metrics_file: outs/fit_metrics.json
 
-train:
-  output_folder: outs/train
-  metrics_folder: outs/train_metrics
-  pretrained_model: MoritzLaurer/xtremedistil-l6-h256-mnli-fever-anli-ling-binary
   epochs: 8
 
-eval:
-  output_folder: outs/eval

src/eval.py

@@ -0,0 +1,35 @@
+import json
+import os
+import pandas as pd
+from collections import Counter
+from pathlib import Path
+
+import fire
+import yaml
+from loguru import logger
+
+
+@logger.catch(reraise=True)
+def fit(data_output_folder, fit_output_folder, fit_metrics_folder):
+    base_dir = os.path.dirname(__file__) + "/.."
+    
+    with open(base_dir+"/params.yaml") as f:
+        params = yaml.safe_load(f)["fit"]
+
+    input_folder = base_dir + "/" + data_output_folder
+    output_folder = base_dir + "/" + fit_output_folder
+    
+    Path(output_folder).mkdir(parents=True, exist_ok=True)
+
+    trades = pd.read_csv(input_folder + "/trades.csv")
+    
+    
+    metrics = {"num_trades": len(trades)}
+    
+    output_metrics_folder = base_dir + "/" + fit_metrics_folder
+    Path(output_metrics_folder).mkdir(parents=True, exist_ok=True)
+    
+    Path(base_dir+"/"+params["metrics_file"]).write_text(json.dumps(metrics, indent=4))
+
+if __name__ == "__main__":
+    fire.Fire(fit)

src/fit.py

@@ -25,7 +25,7 @@ def get_data(output_folder):
 
     logger.info(f"\nSome log.")
 
-    trades = pd.DataFrame(dict(ops_code=["WN COMDTY 1", "CTUSD30Y"], start_time=pd.Timestamp("2022-08-08"), dayfrac=1, quantity=100000, quantity_units="dv01_usd"))
+    trades = pd.DataFrame(dict(ops_code=["WN COMDTY 1", "CTUSD30Y"], start_time=pd.Timestamp(params["data_date"]), dayfrac=1, quantity=100000, quantity_units="dv01_usd"))
 
     trades_file = output_folder + "/trades.csv"
     trades.to_csv(trades_file)