KeyError 500

[l3admin]

I would love to be using AutoTrain / Huggingface for finetuning... but have hit a block.

A couple of us have been attempting to run finetuning, but have hit the KeyError 500, repeatedly. We have watched videos, followed the docs, tried csv, JSONL and many different configs for data formats and Column Mappings... no joy. Even Claude and ChatGPT are stumped.

We are finetuning for a task that involves categorizing (requiring some reasoning, items into categories) so tried SFT and Generic.

It feels like we have missed something obvious and perhaps can not see the wood for the trees - so if anyone can help I would be most grateful!

Example format for Column mapping (we have tried >10 variants in desperation):
{
"input_text": "text",
"category": "label"
}

Example data (again we have tried various formats):
[
{"input_text": "DELL VOSTRO I7 LAPTOP", "category": "Computer Equipment"},
{"input_text": "SAMSUNG 32 FULL HD SMART TV", "category": "Electric, Electronic, Scientific and Digital Equipment (excluding Computers)"},
{"input_text": "TRILITE ALUMINIUM LECTERN", "category": "Display Stands"},
{"input_text": "NOVASTAR MX40 PRO LED PROCESSOR", "category": "Electric, Electronic, Scientific and Digital Equipment (excluding Computers)"},
{"input_text": "NETGEAR 8 PORT 10GB NETWORK SWITCH", "category": "Electric, Electronic, Scientific and Digital Equipment (excluding Computers)"},
{"input_text": "DRUM CYMBAL", "category": "Concert and Musical Equipment – Electric"},
{"input_text": "DW SNARE DRUM", "category": "Concert and Musical Equipment – Electric"}
]

Last few lines from Log (500 KeyError every time):
KeyError: '{ "input_text": "text", "category": "label" }'
raise KeyError(key) from err
File "/app/env/lib/python3.10/site-packages/pandas/core/indexes/base.py", line 3812, in get_loc
indexer = self.columns.get_loc(key)
File "/app/env/lib/python3.10/site-packages/pandas/core/frame.py", line 4102, in getitem
train_df.loc[:, "autotrain_text"] = train_df[self.text_column]

[LYSnow7]

我也遇到了这个问题，应该是列映射的格式有误，但多次修改后仍报错500，不知道为什么