Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,22 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: apache-2.0
|
| 3 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
---
|
| 4 |
+
|
| 5 |
+
## Evo 2
|
| 6 |
+
|
| 7 |
+
Evo 2 is a state-of-the-art DNA language model trained autoregressively.
|
| 8 |
+
|
| 9 |
+
For instructions, details, and examples, please refer to the [github](https://github.com/ArcInstitute/evo2) and [paper]().
|
| 10 |
+
|
| 11 |
+
Evo 2 40B and 7B checkpoints, trained up to 1 million sequence length, are available here:
|
| 12 |
+
| Checkpoint name | Num layers | Num parameters |
|
| 13 |
+
|------------------------------|----|----------|
|
| 14 |
+
| [evo2_40b](https://huggingface.co/facebook/esm2_t48_15B_UR50D) | 50 | 40B |
|
| 15 |
+
| [evo2_7b](https://huggingface.co/facebook/esm2_t36_3B_UR50D) | 32 | 7B |
|
| 16 |
+
|
| 17 |
+
For advanced users, we also share the base models trained on 8192 context length, as well as a 1B checkpoint:
|
| 18 |
+
| Checkpoint name | Num layers | Num parameters |
|
| 19 |
+
|------------------------------|----|----------|
|
| 20 |
+
| [evo2_40b_base](https://huggingface.co/facebook/esm2_t48_15B_UR50D) | 50 | 40B |
|
| 21 |
+
| [evo2_7b_base](https://huggingface.co/facebook/esm2_t36_3B_UR50D) | 32 | 7B |
|
| 22 |
+
| [evo2_1b_base](https://huggingface.co/facebook/esm2_t36_3B_UR50D) | 25 | 1B |
|