Add link to paper
Browse filesThis PR adds a direct link to the paper page for the SalesLLM benchmark, making it easier for users to find the accompanying publication. The model card already contains all relevant metadata and usage instructions; this change only improves citation and discoverability.
README.md
CHANGED
|
@@ -1,10 +1,10 @@
|
|
| 1 |
---
|
| 2 |
-
license: apache-2.0
|
| 3 |
-
library_name: transformers
|
| 4 |
-
pipeline_tag: text-classification
|
| 5 |
language:
|
| 6 |
- zh
|
| 7 |
- en
|
|
|
|
|
|
|
|
|
|
| 8 |
tags:
|
| 9 |
- bert
|
| 10 |
- text-classification
|
|
@@ -16,6 +16,8 @@ tags:
|
|
| 16 |
|
| 17 |
# SaleIntent-BERT
|
| 18 |
|
|
|
|
|
|
|
| 19 |
**SaleIntent-BERT** is a fine-tuned BERT classifier that reads a complete sales conversation and predicts **how the customer ended up** β from clear purchase intent down to hostility. It is the outcome-scoring half of the [SalesLLM benchmark](https://github.com/Bairong-Xdynamics/Benchmarking-LLM-Realistic-Selling-Skill).
|
| 20 |
|
| 21 |
Scoring a sales dialogue with an LLM judge alone conflates two different questions: *did the salesperson run a good process?* and *did the customer actually want to buy at the end?* A model can be articulate, polite, and well-structured while the customer walks away β and an LLM judge, reading the whole transcript, tends to reward the articulate process. SaleIntent-BERT answers the second question independently, by looking only at where the conversation landed.
|
|
@@ -183,4 +185,4 @@ Accuracy is reported over the full 5-class problem. Note that the classes are no
|
|
| 183 |
## Related
|
| 184 |
|
| 185 |
- π [SalesLLM benchmark & code](https://github.com/Bairong-Xdynamics/Benchmarking-LLM-Realistic-Selling-Skill)
|
| 186 |
-
- π€ [CustomerLM](https://huggingface.co/MultiSense/CustomerLM) β the user simulator that generates the customer side of the dialogues this model scores
|
|
|
|
| 1 |
---
|
|
|
|
|
|
|
|
|
|
| 2 |
language:
|
| 3 |
- zh
|
| 4 |
- en
|
| 5 |
+
library_name: transformers
|
| 6 |
+
license: apache-2.0
|
| 7 |
+
pipeline_tag: text-classification
|
| 8 |
tags:
|
| 9 |
- bert
|
| 10 |
- text-classification
|
|
|
|
| 16 |
|
| 17 |
# SaleIntent-BERT
|
| 18 |
|
| 19 |
+
**Paper**: [Sell More, Play Less: Benchmarking LLM Realistic Selling Skill](https://huggingface.co/papers/2604.07054)
|
| 20 |
+
|
| 21 |
**SaleIntent-BERT** is a fine-tuned BERT classifier that reads a complete sales conversation and predicts **how the customer ended up** β from clear purchase intent down to hostility. It is the outcome-scoring half of the [SalesLLM benchmark](https://github.com/Bairong-Xdynamics/Benchmarking-LLM-Realistic-Selling-Skill).
|
| 22 |
|
| 23 |
Scoring a sales dialogue with an LLM judge alone conflates two different questions: *did the salesperson run a good process?* and *did the customer actually want to buy at the end?* A model can be articulate, polite, and well-structured while the customer walks away β and an LLM judge, reading the whole transcript, tends to reward the articulate process. SaleIntent-BERT answers the second question independently, by looking only at where the conversation landed.
|
|
|
|
| 185 |
## Related
|
| 186 |
|
| 187 |
- π [SalesLLM benchmark & code](https://github.com/Bairong-Xdynamics/Benchmarking-LLM-Realistic-Selling-Skill)
|
| 188 |
+
- π€ [CustomerLM](https://huggingface.co/MultiSense/CustomerLM) β the user simulator that generates the customer side of the dialogues this model scores
|