Fix import, argument logic and model mismatch

evaluate/generate.py

@@ -1,11 +1,10 @@
 import os
 import polars as pl
 import pickle
-from tqdm import tqdm
 
-from frieda.providers import (DecoderBase, 
+from providers import (DecoderBase, 
                               init_model, determine_backend)
-from frieda.data_control import (load_data, save_data,
+from data_control import (load_data, save_data,
                                  partition_data)
 
 def respond_q(model: DecoderBase,

evaluate/providers/__init__.py

@@ -1,8 +1,8 @@
 import torch
 import gc
 
-from inference.providers.base import DecoderBase
-from inference.providers.utility import determine_backend
+from evaluate.providers.base import DecoderBase
+from evaluate.providers.utility import determine_backend
 
 # GC and cuda cache empty
 try:
@@ -23,27 +23,27 @@ def init_model(model:str,
                attn_implementation:str,) -> DecoderBase:
     if backend == "hf":
         if 'Qwen/' in model:
-            from inference.providers.hf.qwen import QwenDecoder
+            from evaluate.providers.hf.qwen import QwenDecoder
             return QwenDecoder(model,
                                attn_implementation)
 
         elif 'OpenGVLab/' in model:
-            from inference.providers.hf.internvl import InternVLDecoder
+            from evaluate.providers.hf.internvl import InternVLDecoder
             return InternVLDecoder(model,
                                    attn_implementation)
 
         elif 'AIDC-AI/' in model:
-            from inference.providers.hf.ovis import OvisDecoder
+            from evaluate.providers.hf.ovis import OvisDecoder
             return OvisDecoder(model,
                                attn_implementation)
 
         elif 'llava-hf/' in model:
-            from inference.providers.hf.llava import LLaVADecoder
+            from evaluate.providers.hf.llava import LLaVADecoder
             return LLaVADecoder(model,
                                 attn_implementation)
 
         elif 'zai-org/' in model:
-            from inference.providers.hf.glm import GLMDecoder
+            from evaluate.providers.hf.glm import GLMDecoder
             return GLMDecoder(model,)
 
         else:
@@ -52,15 +52,15 @@ def init_model(model:str,
 
     elif backend == 'openai':
         # Support for GPT family
-        from inference.providers.google import GeminiDecoder
+        from evaluate.providers.openai import GPTDecoder
         print()
 
     elif backend == 'anthropic':
         # Support for Claude family
-        from inference.providers.anthropic import ClaudeDecoder
+        from evaluate.providers.anthropic import ClaudeDecoder
         print()
 
     elif backend == 'google':
         # Supoort for Gemini
-        from inference.providers.openai import GPTDecoder
+        from evaluate.providers.google import GeminiDecoder
         print()

evaluate/providers/anthropic.py

@@ -5,8 +5,8 @@
 
 import torch
 
-from frieda.providers.base import DecoderBase
-from frieda.providers.utility import make_input_message
+from evaluate.providers.base import DecoderBase
+from evaluate.providers.utility import make_input_message
 
 class ClaudeDecoder(DecoderBase):
     def __init__(self,

evaluate/providers/base.py

@@ -1,7 +1,7 @@
 from abc import ABC, abstractmethod
 from typing import List
 
-from frieda.providers.utility import config_bnb
+from evaluate.providers.utility import config_bnb
 
 class DecoderBase(ABC):
     def __init__(self,

evaluate/providers/google.py

@@ -7,8 +7,8 @@
 from google import genai
 from google.genai import types
 
-from frieda.providers.base import DecoderBase
-from frieda.providers.utility import make_input_message
+from evaluate.providers.base import DecoderBase
+from evaluate.providers.utility import make_input_message
 
 class GeminiDecoder(DecoderBase):
     def __init__(self,

evaluate/providers/hf/glm.py

@@ -5,8 +5,8 @@
 import torch
 from transformers import AutoProcessor, Glm4vMoeForConditionalGeneration
 
-from frieda.providers.base import DecoderBase
-from frieda.providers.utility import make_input_message
+from evaluate.providers.base import DecoderBase
+from evaluate.providers.utility import make_input_message
 
 class GLMDecoder(DecoderBase):
     def __init__(self,

evaluate/providers/hf/internvl.py

@@ -5,8 +5,8 @@
 import torch
 from transformers import AutoProcessor
 
-from frieda.providers.base import DecoderBase
-from frieda.providers.utility import make_input_message
+from evaluate.providers.base import DecoderBase
+from evaluate.providers.utility import make_input_message
 
 class InternVLDecoder(DecoderBase):
     def __init__(self,

evaluate/providers/hf/llava.py

@@ -5,8 +5,8 @@
 import torch
 from transformers import AutoProcessor
 
-from frieda.providers.base import DecoderBase
-from frieda.providers.utility import make_input_message
+from evaluate.providers.base import DecoderBase
+from evaluate.providers.utility import make_input_message
 
 class LLaVADecoder(DecoderBase):
     def __init__(self,

evaluate/providers/hf/ovis.py

@@ -5,8 +5,8 @@
 import torch
 from transformers import AutoProcessor, AutoModelForCausalLM
 
-from frieda.providers.base import DecoderBase
-from frieda.providers.utility import make_input_message
+from evaluate.providers.base import DecoderBase
+from evaluate.providers.utility import make_input_message
 
 class OvisDecoder(DecoderBase):
     def __init__(self,

evaluate/providers/hf/qwen.py

@@ -5,8 +5,8 @@
 import torch
 from transformers import AutoProcessor
 
-from frieda.providers.base import DecoderBase
-from frieda.providers.utility import make_input_message
+from evaluate.providers.base import DecoderBase
+from evaluate.providers.utility import make_input_message
 
 class QwenDecoder(DecoderBase):
     def __init__(self,

evaluate/providers/openai.py

@@ -4,8 +4,8 @@
 
 import torch
 
-from frieda.providers.base import DecoderBase
-from frieda.providers.utility import make_input_message
+from evaluate.providers.base import DecoderBase
+from evaluate.providers.utility import make_input_message
 
 class GPTDecoder(DecoderBase):
     def __init__(self,

main.py

@@ -54,34 +54,34 @@ def collect_data(data_dir:str) -> None:
     sub = parser.add_subparsers(dest="cmd", required=True)
 
     p_download = sub.add_parser("download")
-    parser.add_argument("--download_dir", default="./data", 
-                        help="Path to download data to. Default path is ./data")
-    parser.add_arugment("--method", type=str, default='hf', choices=['gdrive', 'hf'],
-                        help="Method of downloading FRIEDA (either Google Drive or HuggingFace)")
+    p_download.add_argument("--download_dir", default="./data",
+                            help="Path to download data to. Default path is ./data")
+    p_download.add_argument("--method", type=str, default='hf', choices=['gdrive', 'hf'],
+                            help="Method of downloading FRIEDA (either Google Drive or HuggingFace)")
 
     p_eval = sub.add_parser("test")
-    parser.add_argument("--model", type=str, default=None,
+    p_eval.add_argument("--model", type=str, required=True,
                         help="Name/model_id/path of LVLM to test on")
-    parser.add_argument("--split", type=str, default='direct', choices=['direct', 'contextual'],
-                        help="FRIEDA evaluation subset (either -direct or -contextual)"),
-    parser.add_argument('--data_dir', type=str, default=None,
-                        help="")
-    parser.add_argument("--result_dir", type=str, default='./results',
+    p_eval.add_argument("--split", type=str, default='direct', choices=['direct', 'contextual'],
+                        help="FRIEDA evaluation subset (either direct or contextual)")
+    p_eval.add_argument('--data_dir', type=str, default=None,
+                        help="Path to the FRIEDA dataset")
+    p_eval.add_argument("--result_dir", type=str, default='./results',
                         help="Path to save evaluation results to. Default path is ./results")
-    parser.add_argument("--batch_size", type=int, default=8,
+    p_eval.add_argument("--batch_size", type=int, default=8,
                         help="Batch size")
-    parser.add_argument("--thinking", action="store_true",
+    p_eval.add_argument("--thinking", action="store_true",
                         help="Run in thinking mode. e.g., Ovis2.5-Thinking")
-    parser.add_argument("--use_flash", action="store_true",
+    p_eval.add_argument("--use_flash", action="store_true",
                         help="Use flash attention 2")
-    parser.add_argument("--evaluate", action="store_true",
+    p_eval.add_argument("--evaluate", action="store_true",
                         help="Run final output evaluation")
 
     # p_visualize = sub.add_parser("visualize")
 
     p_preprocess = sub.add_parser("collection")
-    parser.add_argument("--data_dir", type=str, required=True,
-                        help="Directory with raw PDF files")
+    p_preprocess.add_argument("--data_dir", type=str, required=True,
+                              help="Directory with raw PDF files")
 
     args = parser.parse_args()