multimodalart HF Staff commited on
Commit
f754dbe
·
verified ·
1 Parent(s): 30a9ce4

Map formula prompt key to the equation block type

Browse files
Files changed (1) hide show
  1. app.py +7 -1
app.py CHANGED
@@ -104,6 +104,8 @@ REGION_TASKS = [
104
  ("Chart / scientific figure \u2192 table", "char"),
105
  ("Seal", "seal"),
106
  ]
 
 
107
  # The model prefixes recognized code with its own language marker, e.g. `<_Python_>`.
108
  CODE_LANG_RE = re.compile(r"^\s*<_([A-Za-z0-9+#._\- ]+)_>\s*")
109
 
@@ -297,7 +299,11 @@ def parse_document(
297
  DEFAULT_PROMPTS[task],
298
  _sampling_params(task, max_new_tokens),
299
  )
300
- block = ContentBlock(type=task, bbox=[[0.0, 0.0], [1.0, 1.0]], content=output)
 
 
 
 
301
  blocks = helper.post_process([block]) or [block]
302
  raw_markdown, display_markdown = blocks_to_markdown(page, blocks, False)
303
  seconds = time.time() - started
 
104
  ("Chart / scientific figure \u2192 table", "char"),
105
  ("Seal", "seal"),
106
  ]
107
+ # Prompt keys and block-type names differ for formulas ("formula" vs "equation").
108
+ TASK_BLOCK_TYPES = {"formula": "equation"}
109
  # The model prefixes recognized code with its own language marker, e.g. `<_Python_>`.
110
  CODE_LANG_RE = re.compile(r"^\s*<_([A-Za-z0-9+#._\- ]+)_>\s*")
111
 
 
299
  DEFAULT_PROMPTS[task],
300
  _sampling_params(task, max_new_tokens),
301
  )
302
+ block = ContentBlock(
303
+ type=TASK_BLOCK_TYPES.get(task, task),
304
+ bbox=[[0.0, 0.0], [1.0, 1.0]],
305
+ content=output,
306
+ )
307
  blocks = helper.post_process([block]) or [block]
308
  raw_markdown, display_markdown = blocks_to_markdown(page, blocks, False)
309
  seconds = time.time() - started