Update README.md
Browse files
README.md
CHANGED
|
@@ -1,10 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
# Defog SQLCoder
|
| 2 |
**Updated on Nov 14 to reflect benchmarks for SQLCoder-34B**
|
| 3 |
|
| 4 |
Defog's SQLCoder is a state-of-the-art LLM for converting natural language questions to SQL queries.
|
| 5 |
|
| 6 |
-
[Interactive Demo](https://defog.ai/sqlcoder-demo/) | [🤗 HF Repo](https://huggingface.co/defog/sqlcoder-34b-alpha) | [♾️ Colab](https://colab.research.google.com/drive/1z4rmOEiFkxkMiecAWeTUlPl0OmKgfEu7?usp=sharing) | [🐦 Twitter](https://twitter.com/defogdata)
|
| 7 |
-
|
| 8 |
## TL;DR
|
| 9 |
SQLCoder is a family of large language models that outperforms `gpt-4` and `gpt-4-turbo` for natural language to SQL generation tasks on our [sql-eval](https://github.com/defog-ai/sql-eval) framework, and significantly outperform all popular open-source models.
|
| 10 |
|
|
@@ -66,4 +69,4 @@ SQLCoder-34B has been tested on a 4xA10 GPU with `float16` weights. You can also
|
|
| 66 |
- [x] Open-source the v1 model weights
|
| 67 |
- [x] Train the model on more data, with higher data variance
|
| 68 |
- [ ] Tune the model further with Reward Modelling and RLHF
|
| 69 |
-
- [ ] Pretrain a model from scratch that specializes in SQL analysis
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
tags:
|
| 4 |
+
- code
|
| 5 |
+
---
|
| 6 |
# Defog SQLCoder
|
| 7 |
**Updated on Nov 14 to reflect benchmarks for SQLCoder-34B**
|
| 8 |
|
| 9 |
Defog's SQLCoder is a state-of-the-art LLM for converting natural language questions to SQL queries.
|
| 10 |
|
|
|
|
|
|
|
| 11 |
## TL;DR
|
| 12 |
SQLCoder is a family of large language models that outperforms `gpt-4` and `gpt-4-turbo` for natural language to SQL generation tasks on our [sql-eval](https://github.com/defog-ai/sql-eval) framework, and significantly outperform all popular open-source models.
|
| 13 |
|
|
|
|
| 69 |
- [x] Open-source the v1 model weights
|
| 70 |
- [x] Train the model on more data, with higher data variance
|
| 71 |
- [ ] Tune the model further with Reward Modelling and RLHF
|
| 72 |
+
- [ ] Pretrain a model from scratch that specializes in SQL analysis
|