Update README.md
Browse files
README.md
CHANGED
|
@@ -47,7 +47,7 @@ model-index:
|
|
| 47 |
|
| 48 |
[**🧑💻 Test it on our Demo Space! 🧑💻**](https://huggingface.co/spaces/replit/replit-code-v1-3b-demo)
|
| 49 |
|
| 50 |
-
`replit-code-v1-3b` is a 2.7B Causal Language Model focused on **Code Completion**. The model has been trained on a subset of the Stack Dedup v1.2 dataset.
|
| 51 |
|
| 52 |
The training mixture includes **20 different languages**, listed here in descending order of number of tokens:
|
| 53 |
<br/>
|
|
@@ -139,11 +139,5 @@ Note that as with all code generation models, post-processing of the generated c
|
|
| 139 |
- set `max_tokens` to a reasonable value based on your completion use case
|
| 140 |
- truncate generation to stop words such as `return`, `def`, "```", "`\n\n\n`" to avoid generating incomplete code when `max_tokens` is larger than the length of the expected generated code.
|
| 141 |
|
| 142 |
-
## Inference
|
| 143 |
-
Coming soon.
|
| 144 |
-
|
| 145 |
-
## Evaluation
|
| 146 |
-
Coming soon.
|
| 147 |
-
|
| 148 |
## Model Hash
|
| 149 |
5bc28ce32c6f9aec935ead7b60ea1c46
|
|
|
|
| 47 |
|
| 48 |
[**🧑💻 Test it on our Demo Space! 🧑💻**](https://huggingface.co/spaces/replit/replit-code-v1-3b-demo)
|
| 49 |
|
| 50 |
+
`replit-code-v1-3b` is a 2.7B Causal Language Model focused on **Code Completion**. The model has been trained on a subset of the [Stack Dedup v1.2 dataset](https://arxiv.org/abs/2211.15533).
|
| 51 |
|
| 52 |
The training mixture includes **20 different languages**, listed here in descending order of number of tokens:
|
| 53 |
<br/>
|
|
|
|
| 139 |
- set `max_tokens` to a reasonable value based on your completion use case
|
| 140 |
- truncate generation to stop words such as `return`, `def`, "```", "`\n\n\n`" to avoid generating incomplete code when `max_tokens` is larger than the length of the expected generated code.
|
| 141 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 142 |
## Model Hash
|
| 143 |
5bc28ce32c6f9aec935ead7b60ea1c46
|