יצירת תגים עבור שורות 30000-40000

NHLOCAL · Dec 19, 2024 · fe6f821 · fe6f821
1 parent 18608a1
commit fe6f821
Show file tree

Hide file tree

Showing 4 changed files with 241 additions and 41 deletions.
diff --git a/.github/workflows/run_gemini_v2.yml b/.github/workflows/run_gemini_v2.yml
@@ -8,7 +8,7 @@ on:
       - '.github/workflows/run_gemini_v2.yml'
 
 jobs:
-  process_songs_20001-22000:
+  process_songs_30001-32000:
     runs-on: ubuntu-latest
 
     steps:
@@ -25,19 +25,19 @@ jobs:
           python -m pip install --upgrade pip
           pip install google-generativeai
 
-      - name: Run Python Script for lines 20001-22000
+      - name: Run Python Script for lines 30001-32000
         env:
           GEMINI_API_KEY: ${{ secrets.GEMINI_API_KEY }}
         run: |
-          python machine-learn/scrape_data/level0/Gemini-synthetic-v2/1-gemini_tagging-api.py 20001 22000
+          python machine-learn/scrape_data/level0/Gemini-synthetic-v2/1-gemini_tagging-api.py 30001 32000
 
       - name: Upload Artifact
         uses: actions/upload-artifact@v4
         with:
-          name: song-tags-output-20001-22000
+          name: song-tags-output-30001-32000
           path: tagged_songs.json
 
-  process_songs_22001-24000:
+  process_songs_32001-34000:
     runs-on: ubuntu-latest
 
     steps:
@@ -54,19 +54,19 @@ jobs:
           python -m pip install --upgrade pip
           pip install google-generativeai
 
-      - name: Run Python Script for lines 22001-24000
+      - name: Run Python Script for lines 32001-34000
         env:
           GEMINI_API_KEY: ${{ secrets.GEMINI_API_KEY }}
         run: |
-          python machine-learn/scrape_data/level0/Gemini-synthetic-v2/1-gemini_tagging-api.py 22001 24000
+          python machine-learn/scrape_data/level0/Gemini-synthetic-v2/1-gemini_tagging-api.py 32001 34000
 
       - name: Upload Artifact
         uses: actions/upload-artifact@v4
         with:
-          name: song-tags-output-22001-24000
+          name: song-tags-output-32001-34000
           path: tagged_songs.json
 
-  process_songs_24001-26000:
+  process_songs_34001-36000:
     runs-on: ubuntu-latest
 
     steps:
@@ -83,19 +83,19 @@ jobs:
           python -m pip install --upgrade pip
           pip install google-generativeai
 
-      - name: Run Python Script for lines 24001-26000
+      - name: Run Python Script for lines 34001-36000
         env:
           GEMINI_API_KEY: ${{ secrets.GEMINI_API_KEY }}
         run: |
-          python machine-learn/scrape_data/level0/Gemini-synthetic-v2/1-gemini_tagging-api.py 24001 26000
+          python machine-learn/scrape_data/level0/Gemini-synthetic-v2/1-gemini_tagging-api.py 34001 36000
 
       - name: Upload Artifact
         uses: actions/upload-artifact@v4
         with:
-          name: song-tags-output-24001-26000
+          name: song-tags-output-34001-36000
           path: tagged_songs.json
 
-  process_songs_26001-28000:
+  process_songs_36001-38000:
     runs-on: ubuntu-latest
 
     steps:
@@ -112,19 +112,19 @@ jobs:
           python -m pip install --upgrade pip
           pip install google-generativeai
 
-      - name: Run Python Script for lines 26001-28000
+      - name: Run Python Script for lines 36001-38000
         env:
           GEMINI_API_KEY: ${{ secrets.GEMINI_API_KEY }}
         run: |
-          python machine-learn/scrape_data/level0/Gemini-synthetic-v2/1-gemini_tagging-api.py 26001 28000
+          python machine-learn/scrape_data/level0/Gemini-synthetic-v2/1-gemini_tagging-api.py 36001 38000
 
       - name: Upload Artifact
         uses: actions/upload-artifact@v4
         with:
-          name: song-tags-output-26001-28000
+          name: song-tags-output-36001-38000
           path: tagged_songs.json
 
-  process_songs_28001-30000:
+  process_songs_38001-40000:
     runs-on: ubuntu-latest
 
     steps:
@@ -141,14 +141,14 @@ jobs:
           python -m pip install --upgrade pip
           pip install google-generativeai
 
-      - name: Run Python Script for lines 28001-30000
+      - name: Run Python Script for lines 38001-40000
         env:
           GEMINI_API_KEY: ${{ secrets.GEMINI_API_KEY }}
         run: |
-          python machine-learn/scrape_data/level0/Gemini-synthetic-v2/1-gemini_tagging-api.py 28001 30000
+          python machine-learn/scrape_data/level0/Gemini-synthetic-v2/1-gemini_tagging-api.py 38001 40000
 
       - name: Upload Artifact
         uses: actions/upload-artifact@v4
         with:
-          name: song-tags-output-28001-30000
+          name: song-tags-output-38001-40000
           path: tagged_songs.json
diff --git a/machine-learn/scrape_data/level0/Gemini-synthetic-v2/0-create-yml-list.py b/machine-learn/scrape_data/level0/Gemini-synthetic-v2/0-create-yml-list.py
@@ -40,4 +40,4 @@ def generate_jobs(start, end, step, output_file):
         f.write("\n".join(jobs))
 
 # Generate jobs for ranges 1000 to 5000 with a step of 1000
-generate_jobs(20001, 30000, 2000, 'jobs_output.txt')
+generate_jobs(30001, 40000, 2000, 'jobs_output.txt')
diff --git a/machine-learn/scrape_data/level0/Gemini-synthetic-v2/iteration_data.json b/machine-learn/scrape_data/level0/Gemini-synthetic-v2/iteration_data.json
@@ -0,0 +1,200 @@
+{
+  "0": {
+    "ner": 32594.164313317735
+  },
+  "1": {
+    "ner": 20399.96439350402
+  },
+  "2": {
+    "ner": 16058.678328589409
+  },
+  "3": {
+    "ner": 14119.626520210864
+  },
+  "4": {
+    "ner": 13400.467329233605
+  },
+  "5": {
+    "ner": 12631.384810704743
+  },
+  "6": {
+    "ner": 12240.227350688649
+  },
+  "7": {
+    "ner": 11768.314378044452
+  },
+  "8": {
+    "ner": 11607.811493621062
+  },
+  "9": {
+    "ner": 11218.239474557216
+  },
+  "10": {
+    "ner": 11037.596818363272
+  },
+  "11": {
+    "ner": 11063.021051025255
+  },
+  "12": {
+    "ner": 10457.756479986085
+  },
+  "13": {
+    "ner": 10532.406894666514
+  },
+  "14": {
+    "ner": 10227.484260950203
+  },
+  "15": {
+    "ner": 10215.15850648792
+  },
+  "16": {
+    "ner": 10010.11222164606
+  },
+  "17": {
+    "ner": 9941.071052523734
+  },
+  "18": {
+    "ner": 9811.91049791307
+  },
+  "19": {
+    "ner": 9455.067349402849
+  },
+  "20": {
+    "ner": 9323.480768392976
+  },
+  "21": {
+    "ner": 9243.659595690471
+  },
+  "22": {
+    "ner": 9250.385300274715
+  },
+  "23": {
+    "ner": 9129.943736823096
+  },
+  "24": {
+    "ner": 8997.397397997704
+  },
+  "25": {
+    "ner": 8781.381484326543
+  },
+  "26": {
+    "ner": 8799.50681115516
+  },
+  "27": {
+    "ner": 8768.027454169775
+  },
+  "28": {
+    "ner": 8680.024729407478
+  },
+  "29": {
+    "ner": 8485.69218599471
+  },
+  "30": {
+    "ner": 8459.001738092167
+  },
+  "31": {
+    "ner": 8447.894452833027
+  },
+  "32": {
+    "ner": 8412.43210193547
+  },
+  "33": {
+    "ner": 8247.87361789489
+  },
+  "34": {
+    "ner": 8076.676813265454
+  },
+  "35": {
+    "ner": 7983.367212751617
+  },
+  "36": {
+    "ner": 8047.692774950663
+  },
+  "37": {
+    "ner": 7876.0155811355025
+  },
+  "38": {
+    "ner": 7974.245524379026
+  },
+  "39": {
+    "ner": 7899.06225185776
+  },
+  "40": {
+    "ner": 7936.552281323195
+  },
+  "41": {
+    "ner": 7804.695917784534
+  },
+  "42": {
+    "ner": 7751.268585785002
+  },
+  "43": {
+    "ner": 7630.777629555615
+  },
+  "44": {
+    "ner": 7557.969069023306
+  },
+  "45": {
+    "ner": 7516.585152319071
+  },
+  "46": {
+    "ner": 7349.033556689978
+  },
+  "47": {
+    "ner": 7409.912806656053
+  },
+  "48": {
+    "ner": 7301.788914962628
+  },
+  "49": {
+    "ner": 7273.647090282899
+  },
+  "50": {
+    "ner": 7427.05447019191
+  },
+  "51": {
+    "ner": 7296.00502661919
+  },
+  "52": {
+    "ner": 7329.644561921469
+  },
+  "53": {
+    "ner": 7233.506164490878
+  },
+  "54": {
+    "ner": 7005.504783939925
+  },
+  "55": {
+    "ner": 7141.23487458332
+  },
+  "56": {
+    "ner": 6905.3384089048695
+  },
+  "57": {
+    "ner": 6925.590931135269
+  },
+  "58": {
+    "ner": 6960.05837120723
+  },
+  "59": {
+    "ner": 6915.8124689869
+  },
+  "60": {
+    "ner": 6700.680239110464
+  },
+  "61": {
+    "ner": 6811.550214059638
+  },
+  "62": {
+    "ner": 6706.345974776895
+  },
+  "63": {
+    "ner": 6884.8719234185955
+  },
+  "64": {
+    "ner": 6753.397600133148
+  },
+  "65": {
+    "ner": 6768.124060673756
+  }
+}