I’m finding that the assistants api works ok when retrieving small files. But if I give it a 5mb json file containing 1000 blogs posts, my runs all expire.
I’m experiencing this in the playground, not even trying to use the API yet.
Example run below. This same run executes fine if the provided file is more like 50kb:
"object": "list",
"data": [
{
"id": "run_WG5mndIJ9Cd3qNK1Nq65uL0F",
"object": "thread.run",
"created_at": 1711029855,
"assistant_id": "asst_by3qpL6IO5SQq5OCrJkZYIwq",
"thread_id": "thread_172WH58K6BDCKkq9RtebIH8e",
"status": "expired",
"started_at": 1711029856,
"expires_at": null,
"cancelled_at": null,
"failed_at": null,
"completed_at": null,
"required_action": null,
"last_error": null,
"model": "gpt-4-turbo-preview",
"instructions": "you are an attorney and legal blogger.\n\nuse the contents of the attached file to inform your response. it is an array of blog posts in json format.\n\nafter your response, provide a json array of which blog posts you used for your response. include the title and slug. delineate this section by wrapping it in backticks and the word json like this:\n\n```json\n{\n{\n\"title\": \"example blog post\",\n\"slug\": \"example-blog-post\",\n},\n{\n\"title\": \"other example blog post\",\n\"slug\": \"123\",\n}\n\n}\n```\n\ndo not preface the footnotes section by saying \"footnotes\" or anything like that. I'm going to extract it programmatically. \nuse the contents of the attached file to inform your response. it is an array of blog posts in json format.\n\nafter your response, provide a json array of which blog posts you used for your response. include the title and slug. delineate this section by wrapping it in backticks and the word json like this:\n\n```json\n{\n{\n\"title\": \"example blog post\",\n\"slug\": \"example-blog-post\",\n},\n{\n\"title\": \"other example blog post\",\n\"slug\": \"123\",\n}\n\n}\n```\n\ndo not preface the footnotes section by saying \"footnotes\" or anything like that. I'm going to extract it programmatically.",
"tools": [
{
"type": "retrieval"
}
],
"file_ids": [
"file-laAme9ZZJwBg4SoZf30o4KMk"
],
"metadata": {},
"usage": {
"prompt_tokens": 754,
"completion_tokens": 29,
"total_tokens": 783
}
}
],
"first_id": "run_WG5mndIJ9Cd3qNK1Nq65uL0F",
"last_id": "run_WG5mndIJ9Cd3qNK1Nq65uL0F",
"has_more": false
}```