|
|
|
@ -7,7 +7,7 @@
|
|
|
|
|
"source": [
|
|
|
|
|
"# Structured Decoding with RELLM\n",
|
|
|
|
|
"\n",
|
|
|
|
|
"[RELLM](https://github.com/r2d4/rellm) is a library that wraps local HuggingFace pipeline models for structured decoding.\n",
|
|
|
|
|
"[RELLM](https://github.com/r2d4/rellm) is a library that wraps local Hugging Face pipeline models for structured decoding.\n",
|
|
|
|
|
"\n",
|
|
|
|
|
"It works by generating tokens one at a time. At each step, it masks tokens that don't conform to the provided partial regular expression.\n",
|
|
|
|
|
"\n",
|
|
|
|
@ -32,7 +32,7 @@
|
|
|
|
|
"id": "66bd89f1-8daa-433d-bb8f-5b0b3ae34b00",
|
|
|
|
|
"metadata": {},
|
|
|
|
|
"source": [
|
|
|
|
|
"### HuggingFace Baseline\n",
|
|
|
|
|
"### Hugging Face Baseline\n",
|
|
|
|
|
"\n",
|
|
|
|
|
"First, let's establish a qualitative baseline by checking the output of the model without structured decoding."
|
|
|
|
|
]
|
|
|
|
|