add handler.py for HF Dedicated Inference

This PR adds a [Inference Handler](https://huggingface.co/docs/inference-endpoints/guides/custom_handler) to this model.
This is required for using it in a HuggingFace Dedicated Endpoint, since this product does not has a Image-To-Text task available out-of-the-box.

The handler implements the inference as described in the (model docs)[https://huggingface.co/docs/transformers/model_doc/nougat] but with the StoppingCriteria implemented on the notebook example.

handler.py

@@ -0,0 +1,102 @@
+from transformers import NougatProcessor, VisionEncoderDecoderModel, StoppingCriteria, StoppingCriteriaList
+import torch.cuda
+import io
+import base64
+from PIL import Image
+from typing import Dict, Any
+from collections import defaultdict
+
+class RunningVarTorch:
+  def __init__(self, L=15, norm=False):
+    self.values = None
+    self.L = L
+    self.norm = norm
+
+  def push(self, x: torch.Tensor):
+    assert x.dim() == 1
+    if self.values is None:
+      self.values = x[:, None]
+    elif self.values.shape[1] < self.L:
+      self.values = torch.cat((self.values, x[:, None]), 1)
+    else:
+      self.values = torch.cat((self.values[:, 1:], x[:, None]), 1)
+
+  def variance(self):
+    if self.values is None:
+      return
+    if self.norm:
+      return torch.var(self.values, 1) / self.values.shape[1]
+    else:
+      return torch.var(self.values, 1)
+
+class StoppingCriteriaScores(StoppingCriteria):
+  def __init__(self, threshold: float = 0.015, window_size: int = 200):
+    super().__init__()
+    self.threshold = threshold
+    self.vars = RunningVarTorch(norm=True)
+    self.varvars = RunningVarTorch(L=window_size)
+    self.stop_inds = defaultdict(int)
+    self.stopped = defaultdict(bool)
+    self.size = 0
+    self.window_size = window_size
+
+  @torch.no_grad()
+  def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor):
+    last_scores = scores[-1]
+    self.vars.push(last_scores.max(1)[0].float().cpu())
+    self.varvars.push(self.vars.variance())
+    self.size += 1
+    if self.size < self.window_size:
+      return False
+
+    varvar = self.varvars.variance()
+    for b in range(len(last_scores)):
+      if varvar[b] < self.threshold:
+        if self.stop_inds[b] > 0 and not self.stopped[b]:
+          self.stopped[b] = self.stop_inds[b] >= self.size
+        else:
+          self.stop_inds[b] = int(
+              min(max(self.size, 1) * 1.15 + 150 + self.window_size, 4095)
+          )
+      else:
+        self.stop_inds[b] = 0
+        self.stopped[b] = False
+    return all(self.stopped.values()) and len(self.stopped) > 0
+
+class EndpointHandler():
+  def __init__(self, path="facebook/nougat-base"):
+    self.device = "cuda" if torch.cuda.is_available() else "cpu"
+    self.processor = NougatProcessor.from_pretrained(path)
+    self.model = VisionEncoderDecoderModel.from_pretrained(path)
+    self.model = self.model.to(self.device)
+
+  def __call__(self, data: Dict[str, Any]) -> Dict[str, Any]:
+    """
+    Args:
+      data (Dict): The payload with the text prompt 
+    and generation parameters.
+    """
+    # Get inputs
+    input = data.pop("inputs", None)
+    parameters = data.pop("parameters", None)
+    fix_markdown = data.pop("fix_markdown", None)
+    if input is None:
+      raise ValueError("Missing image.")
+    # autoregressively generate tokens, with custom stopping criteria (as defined by the Nougat authors)
+    binary_data = base64.b64decode(input)
+
+    image = Image.open(io.BytesIO(binary_data))
+    pixel_values = self.processor(images= image, return_tensors="pt").pixel_values
+    outputs = self.model.generate(
+      pixel_values=pixel_values.to(self.model.device),
+      min_length=1,
+      bad_words_ids=[[self.processor.tokenizer.unk_token_id]],
+      return_dict_in_generate=True,
+      output_scores=True,
+      stopping_criteria=StoppingCriteriaList([StoppingCriteriaScores()]),
+      **parameters,
+    )
+    generated = self.processor.batch_decode(outputs[0], skip_special_tokens=True)[0]
+    prediction = self.processor.post_process_generation(generated, fix_markdown=fix_markdown)
+
+    return {"generated_text": prediction}

requirements.txt

@@ -0,0 +1,2 @@
+python-Levenshtein
+nltk