Search Test Runs

Test T0284 at 2026-01-24

{'document-type': ['manuscript'], 'writing': ['handwritten'], 'century': [15], 'language': ['de'], 'layout': ['prose'], 'task': ['transcription']}

IDENTITY and PURPOSE:

You are presented with an image from a 15th century medieval manuscript written in Basel in late medieval German. Your task is to extract the text from the manuscript in the specified JSON format. You must extract all text exactly as it appears in the manuscript, maintaining historical spellings, punctuation, and formatting. Do not resolve abbreviations.

The manuscript may contain:
1. A main text body (often in one column as continuous text)
2. A folio number (if present)
3. Additional notes or text written in the margins (labeled as addition1, addition2, etc.)

You must:
- Identify and transcribe the main text
- Extract the folio number if visible (use empty string "" if not visible)
- Identify and transcribe any marginal additions separately
- Preserve line breaks with \n
- Maintain all historical spellings and abbreviations exactly as written. If a letter is superscribed, normalize it by writing the superscribed letter after the main letter, e.g. "u with superscribed o" is spelled as "uo". If special characters are used for abbreviations, do not resolve them, but try to transcribe the special character according to the Medieval Unicode Font Initiative. ꝛ for "er" and ꝰ for "us" or "em" might be the most common special characters.
- Do not modernize or correct the text
- Do not use OCR or attempt to interpret unclear text - transcribe what you can see
- If a field has no content, use an empty string "" (not null)

Take a deep breath and think step by step about the layout of the page. First identify the folio number, then the main text area, then any marginal additions. Return only a JSON file with no additional commentary.

EXAMPLE OUTPUT:

{
  "folios": [
    {
      "folio": "15",
      "text": "In disem jare kam der kunig\n mit grossem here in daz lant\n vnd belagerte die stat Basel\n do wertend sich die burger\n mit grosser kraft vnd tugent\n vnd triben den kunig dannen\n mit schanden vnd verlust",
      "addition1": "Anno domini 1444",
      "addition2": "",
      "addition3": ""
    }
  ]
}

no valid result

Test T0272 at 2026-01-24

{'document-type': ['manuscript'], 'writing': ['handwritten'], 'century': [15], 'language': ['de'], 'layout': ['prose'], 'task': ['transcription']}

IDENTITY and PURPOSE:

You are presented with an image from a 15th century medieval manuscript written in Basel in late medieval German. Your task is to extract the text from the manuscript in the specified JSON format. You must extract all text exactly as it appears in the manuscript, maintaining historical spellings, punctuation, and formatting. Do not resolve abbreviations.

The manuscript may contain:
1. A main text body (often in one column as continuous text)
2. A folio number (if present)
3. Additional notes or text written in the margins (labeled as addition1, addition2, etc.)

You must:
- Identify and transcribe the main text
- Extract the folio number if visible (use empty string "" if not visible)
- Identify and transcribe any marginal additions separately
- Preserve line breaks with \n
- Maintain all historical spellings and abbreviations exactly as written. If a letter is superscribed, normalize it by writing the superscribed letter after the main letter, e.g. "u with superscribed o" is spelled as "uo". If special characters are used for abbreviations, do not resolve them, but try to transcribe the special character according to the Medieval Unicode Font Initiative. ꝛ for "er" and ꝰ for "us" or "em" might be the most common special characters.
- Do not modernize or correct the text
- Do not use OCR or attempt to interpret unclear text - transcribe what you can see
- If a field has no content, use an empty string "" (not null)

Take a deep breath and think step by step about the layout of the page. First identify the folio number, then the main text area, then any marginal additions. Return only a JSON file with no additional commentary.

EXAMPLE OUTPUT:

{
  "folios": [
    {
      "folio": "15",
      "text": "In disem jare kam der kunig\n mit grossem here in daz lant\n vnd belagerte die stat Basel\n do wertend sich die burger\n mit grosser kraft vnd tugent\n vnd triben den kunig dannen\n mit schanden vnd verlust",
      "addition1": "Anno domini 1444",
      "addition2": "",
      "addition3": ""
    }
  ]
}

no valid result

Test T0293 at 2026-01-24

{'document-type': ['manuscript'], 'writing': ['handwritten'], 'century': [15], 'language': ['de'], 'layout': ['prose'], 'task': ['transcription']}

IDENTITY and PURPOSE:

You are presented with an image from a 15th century medieval manuscript written in Basel in late medieval German. Your task is to extract the text from the manuscript in the specified JSON format. You must extract all text exactly as it appears in the manuscript, maintaining historical spellings, punctuation, and formatting. Do not resolve abbreviations.

The manuscript may contain:
1. A main text body (often in one column as continuous text)
2. A folio number (if present)
3. Additional notes or text written in the margins (labeled as addition1, addition2, etc.)

You must:
- Identify and transcribe the main text
- Extract the folio number if visible (use empty string "" if not visible)
- Identify and transcribe any marginal additions separately
- Preserve line breaks with \n
- Maintain all historical spellings and abbreviations exactly as written. If a letter is superscribed, normalize it by writing the superscribed letter after the main letter, e.g. "u with superscribed o" is spelled as "uo". If special characters are used for abbreviations, do not resolve them, but try to transcribe the special character according to the Medieval Unicode Font Initiative. ꝛ for "er" and ꝰ for "us" or "em" might be the most common special characters.
- Do not modernize or correct the text
- Do not use OCR or attempt to interpret unclear text - transcribe what you can see
- If a field has no content, use an empty string "" (not null)

Take a deep breath and think step by step about the layout of the page. First identify the folio number, then the main text area, then any marginal additions. Return only a JSON file with no additional commentary.

EXAMPLE OUTPUT:

{
  "folios": [
    {
      "folio": "15",
      "text": "In disem jare kam der kunig\n mit grossem here in daz lant\n vnd belagerte die stat Basel\n do wertend sich die burger\n mit grosser kraft vnd tugent\n vnd triben den kunig dannen\n mit schanden vnd verlust",
      "addition1": "Anno domini 1444",
      "addition2": "",
      "addition3": ""
    }
  ]
}

no valid result

Test T0441 at 2025-11-25

{'document-type': ['manuscript'], 'writing': ['handwritten'], 'century': [15], 'language': ['de'], 'layout': ['prose'], 'task': ['transcription']}

IDENTITY and PURPOSE:

You are presented with an image from a 15th century medieval manuscript written in Basel in late medieval German. Your task is to extract the text from the manuscript in the specified JSON format. You must extract all text exactly as it appears in the manuscript, maintaining historical spellings, punctuation, and formatting. Do not resolve abbreviations.

The manuscript may contain:
1. A main text body (often in one column as continuous text)
2. A folio number (if present)
3. Additional notes or text written in the margins (labeled as addition1, addition2, etc.)

You must:
- Identify and transcribe the main text
- Extract the folio number if visible (use empty string "" if not visible)
- Identify and transcribe any marginal additions separately
- Preserve line breaks with \n
- Maintain all historical spellings and abbreviations exactly as written. If a letter is superscribed, normalize it by writing the superscribed letter after the main letter, e.g. "u with superscribed o" is spelled as "uo". If special characters are used for abbreviations, do not resolve them, but try to transcribe the special character according to the Medieval Unicode Font Initiative. ꝛ for "er" and ꝰ for "us" or "em" might be the most common special characters.
- Do not modernize or correct the text
- Do not use OCR or attempt to interpret unclear text - transcribe what you can see
- If a field has no content, use an empty string "" (not null)

Take a deep breath and think step by step about the layout of the page. First identify the folio number, then the main text area, then any marginal additions. Return only a JSON file with no additional commentary.

EXAMPLE OUTPUT:

{
  "folios": [
    {
      "folio": "15",
      "text": "In disem jare kam der kunig\n mit grossem here in daz lant\n vnd belagerte die stat Basel\n do wertend sich die burger\n mit grosser kraft vnd tugent\n vnd triben den kunig dannen\n mit schanden vnd verlust",
      "addition1": "Anno domini 1444",
      "addition2": "",
      "addition3": ""
    }
  ]
}

no valid result

Test T0409 at 2025-11-25

{'document-type': ['manuscript'], 'writing': ['handwritten'], 'century': [15], 'language': ['de'], 'layout': ['prose'], 'task': ['transcription']}

IDENTITY and PURPOSE:

You are presented with an image from a 15th century medieval manuscript written in Basel in late medieval German. Your task is to extract the text from the manuscript in the specified JSON format. You must extract all text exactly as it appears in the manuscript, maintaining historical spellings, punctuation, and formatting. Do not resolve abbreviations.

The manuscript may contain:
1. A main text body (often in one column as continuous text)
2. A folio number (if present)
3. Additional notes or text written in the margins (labeled as addition1, addition2, etc.)

You must:
- Identify and transcribe the main text
- Extract the folio number if visible (use empty string "" if not visible)
- Identify and transcribe any marginal additions separately
- Preserve line breaks with \n
- Maintain all historical spellings and abbreviations exactly as written. If a letter is superscribed, normalize it by writing the superscribed letter after the main letter, e.g. "u with superscribed o" is spelled as "uo". If special characters are used for abbreviations, do not resolve them, but try to transcribe the special character according to the Medieval Unicode Font Initiative. ꝛ for "er" and ꝰ for "us" or "em" might be the most common special characters.
- Do not modernize or correct the text
- Do not use OCR or attempt to interpret unclear text - transcribe what you can see
- If a field has no content, use an empty string "" (not null)

Take a deep breath and think step by step about the layout of the page. First identify the folio number, then the main text area, then any marginal additions. Return only a JSON file with no additional commentary.

EXAMPLE OUTPUT:

{
  "folios": [
    {
      "folio": "15",
      "text": "In disem jare kam der kunig\n mit grossem here in daz lant\n vnd belagerte die stat Basel\n do wertend sich die burger\n mit grosser kraft vnd tugent\n vnd triben den kunig dannen\n mit schanden vnd verlust",
      "addition1": "Anno domini 1444",
      "addition2": "",
      "addition3": ""
    }
  ]
}

no valid result

Test T0271 at 2025-11-24

{'document-type': ['manuscript'], 'writing': ['handwritten'], 'century': [15], 'language': ['de'], 'layout': ['prose'], 'task': ['transcription']}

IDENTITY and PURPOSE:

You are presented with an image from a 15th century medieval manuscript written in Basel in late medieval German. Your task is to extract the text from the manuscript in the specified JSON format. You must extract all text exactly as it appears in the manuscript, maintaining historical spellings, punctuation, and formatting. Do not resolve abbreviations.

The manuscript may contain:
1. A main text body (often in one column as continuous text)
2. A folio number (if present)
3. Additional notes or text written in the margins (labeled as addition1, addition2, etc.)

You must:
- Identify and transcribe the main text
- Extract the folio number if visible (use empty string "" if not visible)
- Identify and transcribe any marginal additions separately
- Preserve line breaks with \n
- Maintain all historical spellings and abbreviations exactly as written. If a letter is superscribed, normalize it by writing the superscribed letter after the main letter, e.g. "u with superscribed o" is spelled as "uo". If special characters are used for abbreviations, do not resolve them, but try to transcribe the special character according to the Medieval Unicode Font Initiative. ꝛ for "er" and ꝰ for "us" or "em" might be the most common special characters.
- Do not modernize or correct the text
- Do not use OCR or attempt to interpret unclear text - transcribe what you can see
- If a field has no content, use an empty string "" (not null)

Take a deep breath and think step by step about the layout of the page. First identify the folio number, then the main text area, then any marginal additions. Return only a JSON file with no additional commentary.

EXAMPLE OUTPUT:

{
  "folios": [
    {
      "folio": "15",
      "text": "In disem jare kam der kunig\n mit grossem here in daz lant\n vnd belagerte die stat Basel\n do wertend sich die burger\n mit grosser kraft vnd tugent\n vnd triben den kunig dannen\n mit schanden vnd verlust",
      "addition1": "Anno domini 1444",
      "addition2": "",
      "addition3": ""
    }
  ]
}

no valid result

Test T0298 at 2025-10-24

{'document-type': ['manuscript'], 'writing': ['handwritten'], 'century': [15], 'language': ['de'], 'layout': ['prose'], 'task': ['transcription']}

IDENTITY and PURPOSE:

You are presented with an image from a 15th century medieval manuscript written in Basel in late medieval German. Your task is to extract the text from the manuscript in the specified JSON format. You must extract all text exactly as it appears in the manuscript, maintaining historical spellings, punctuation, and formatting. Do not resolve abbreviations.

The manuscript may contain:
1. A main text body (often in one column as continuous text)
2. A folio number (if present)
3. Additional notes or text written in the margins (labeled as addition1, addition2, etc.)

You must:
- Identify and transcribe the main text
- Extract the folio number if visible (use empty string "" if not visible)
- Identify and transcribe any marginal additions separately
- Preserve line breaks with \n
- Maintain all historical spellings and abbreviations exactly as written. If a letter is superscribed, normalize it by writing the superscribed letter after the main letter, e.g. "u with superscribed o" is spelled as "uo". If special characters are used for abbreviations, do not resolve them, but try to transcribe the special character according to the Medieval Unicode Font Initiative. ꝛ for "er" and ꝰ for "us" or "em" might be the most common special characters.
- Do not modernize or correct the text
- Do not use OCR or attempt to interpret unclear text - transcribe what you can see
- If a field has no content, use an empty string "" (not null)

Take a deep breath and think step by step about the layout of the page. First identify the folio number, then the main text area, then any marginal additions. Return only a JSON file with no additional commentary.

EXAMPLE OUTPUT:

{
  "folios": [
    {
      "folio": "15",
      "text": "In disem jare kam der kunig\n mit grossem here in daz lant\n vnd belagerte die stat Basel\n do wertend sich die burger\n mit grosser kraft vnd tugent\n vnd triben den kunig dannen\n mit schanden vnd verlust",
      "addition1": "Anno domini 1444",
      "addition2": "",
      "addition3": ""
    }
  ]
}

no valid result

Test T0278 at 2025-10-24

{'document-type': ['manuscript'], 'writing': ['handwritten'], 'century': [15], 'language': ['de'], 'layout': ['prose'], 'task': ['transcription']}

IDENTITY and PURPOSE:

You are presented with an image from a 15th century medieval manuscript written in Basel in late medieval German. Your task is to extract the text from the manuscript in the specified JSON format. You must extract all text exactly as it appears in the manuscript, maintaining historical spellings, punctuation, and formatting. Do not resolve abbreviations.

The manuscript may contain:
1. A main text body (often in one column as continuous text)
2. A folio number (if present)
3. Additional notes or text written in the margins (labeled as addition1, addition2, etc.)

You must:
- Identify and transcribe the main text
- Extract the folio number if visible (use empty string "" if not visible)
- Identify and transcribe any marginal additions separately
- Preserve line breaks with \n
- Maintain all historical spellings and abbreviations exactly as written. If a letter is superscribed, normalize it by writing the superscribed letter after the main letter, e.g. "u with superscribed o" is spelled as "uo". If special characters are used for abbreviations, do not resolve them, but try to transcribe the special character according to the Medieval Unicode Font Initiative. ꝛ for "er" and ꝰ for "us" or "em" might be the most common special characters.
- Do not modernize or correct the text
- Do not use OCR or attempt to interpret unclear text - transcribe what you can see
- If a field has no content, use an empty string "" (not null)

Take a deep breath and think step by step about the layout of the page. First identify the folio number, then the main text area, then any marginal additions. Return only a JSON file with no additional commentary.

EXAMPLE OUTPUT:

{
  "folios": [
    {
      "folio": "15",
      "text": "In disem jare kam der kunig\n mit grossem here in daz lant\n vnd belagerte die stat Basel\n do wertend sich die burger\n mit grosser kraft vnd tugent\n vnd triben den kunig dannen\n mit schanden vnd verlust",
      "addition1": "Anno domini 1444",
      "addition2": "",
      "addition3": ""
    }
  ]
}

no valid result

Test T0295 at 2025-10-24

{'document-type': ['manuscript'], 'writing': ['handwritten'], 'century': [15], 'language': ['de'], 'layout': ['prose'], 'task': ['transcription']}

IDENTITY and PURPOSE:

You are presented with an image from a 15th century medieval manuscript written in Basel in late medieval German. Your task is to extract the text from the manuscript in the specified JSON format. You must extract all text exactly as it appears in the manuscript, maintaining historical spellings, punctuation, and formatting. Do not resolve abbreviations.

The manuscript may contain:
1. A main text body (often in one column as continuous text)
2. A folio number (if present)
3. Additional notes or text written in the margins (labeled as addition1, addition2, etc.)

You must:
- Identify and transcribe the main text
- Extract the folio number if visible (use empty string "" if not visible)
- Identify and transcribe any marginal additions separately
- Preserve line breaks with \n
- Maintain all historical spellings and abbreviations exactly as written. If a letter is superscribed, normalize it by writing the superscribed letter after the main letter, e.g. "u with superscribed o" is spelled as "uo". If special characters are used for abbreviations, do not resolve them, but try to transcribe the special character according to the Medieval Unicode Font Initiative. ꝛ for "er" and ꝰ for "us" or "em" might be the most common special characters.
- Do not modernize or correct the text
- Do not use OCR or attempt to interpret unclear text - transcribe what you can see
- If a field has no content, use an empty string "" (not null)

Take a deep breath and think step by step about the layout of the page. First identify the folio number, then the main text area, then any marginal additions. Return only a JSON file with no additional commentary.

EXAMPLE OUTPUT:

{
  "folios": [
    {
      "folio": "15",
      "text": "In disem jare kam der kunig\n mit grossem here in daz lant\n vnd belagerte die stat Basel\n do wertend sich die burger\n mit grosser kraft vnd tugent\n vnd triben den kunig dannen\n mit schanden vnd verlust",
      "addition1": "Anno domini 1444",
      "addition2": "",
      "addition3": ""
    }
  ]
}

no valid result

Test T0281 at 2025-10-24

{'document-type': ['manuscript'], 'writing': ['handwritten'], 'century': [15], 'language': ['de'], 'layout': ['prose'], 'task': ['transcription']}

IDENTITY and PURPOSE:

You are presented with an image from a 15th century medieval manuscript written in Basel in late medieval German. Your task is to extract the text from the manuscript in the specified JSON format. You must extract all text exactly as it appears in the manuscript, maintaining historical spellings, punctuation, and formatting. Do not resolve abbreviations.

The manuscript may contain:
1. A main text body (often in one column as continuous text)
2. A folio number (if present)
3. Additional notes or text written in the margins (labeled as addition1, addition2, etc.)

You must:
- Identify and transcribe the main text
- Extract the folio number if visible (use empty string "" if not visible)
- Identify and transcribe any marginal additions separately
- Preserve line breaks with \n
- Maintain all historical spellings and abbreviations exactly as written. If a letter is superscribed, normalize it by writing the superscribed letter after the main letter, e.g. "u with superscribed o" is spelled as "uo". If special characters are used for abbreviations, do not resolve them, but try to transcribe the special character according to the Medieval Unicode Font Initiative. ꝛ for "er" and ꝰ for "us" or "em" might be the most common special characters.
- Do not modernize or correct the text
- Do not use OCR or attempt to interpret unclear text - transcribe what you can see
- If a field has no content, use an empty string "" (not null)

Take a deep breath and think step by step about the layout of the page. First identify the folio number, then the main text area, then any marginal additions. Return only a JSON file with no additional commentary.

EXAMPLE OUTPUT:

{
  "folios": [
    {
      "folio": "15",
      "text": "In disem jare kam der kunig\n mit grossem here in daz lant\n vnd belagerte die stat Basel\n do wertend sich die burger\n mit grosser kraft vnd tugent\n vnd triben den kunig dannen\n mit schanden vnd verlust",
      "addition1": "Anno domini 1444",
      "addition2": "",
      "addition3": ""
    }
  ]
}

no valid result