Update README.md
Browse files
README.md
CHANGED
|
@@ -7,16 +7,8 @@ library_name: transformers
|
|
| 7 |
|
| 8 |
|
| 9 |
|
| 10 |
-
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
Linear models offer a promising approach to significantly reduce computational costs at scale, particularly for large context lengths. Enabling a >1000x improvement in inference costs, enabling o1 inference time thinking and wider AI accessibility.
|
| 14 |
-
|
| 15 |
-
As demonstrated with our Qwerky-72B-Preview and prior models such as QRWKV6-32B Instruct Preview, we have successfully converted Qwen 2.5 72B into a RWKV variant without requiring a pretrain on the base model or retraining the model from scratch. Enabling us to test and validate the more efficient RWKV Linear attention with a much smaller budget. Since our preview, we have continued to refine our technique and managed to improve the model over the preview model iteration.
|
| 16 |
-
|
| 17 |
-
As with our previous models, the model's inherent knowledge and dataset training are inherited from its "parent" model. Consequently, unlike previous RWKV models trained on over 100+ languages, the QRWKV model is limited to approximately 30 languages supported by the Qwen line of models.
|
| 18 |
-
|
| 19 |
-
You may find our details of the process from our previous release, [here](https://huggingface.co/recursal/QRWKV6-32B-Instruct-Preview-v0.1).
|
| 20 |
|
| 21 |
Benchmarks is as follows for both Qwerky-QwQ-32B and Qwerky-72B models:
|
| 22 |
|
|
@@ -80,4 +72,14 @@ generated_ids = [
|
|
| 80 |
]
|
| 81 |
|
| 82 |
response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
|
| 83 |
-
```
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7 |
|
| 8 |
|
| 9 |
|
| 10 |
+
- Try out the model on [](https://featherless.ai/models/featherless-ai/Qwerky-72B)
|
| 11 |
+
- Model details can be found on our blog post here! [](https://substack.recursal.ai/p/qwerky-72b-and-32b-training-large)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
|
| 13 |
Benchmarks is as follows for both Qwerky-QwQ-32B and Qwerky-72B models:
|
| 14 |
|
|
|
|
| 72 |
]
|
| 73 |
|
| 74 |
response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
|
| 75 |
+
```
|
| 76 |
+
|
| 77 |
+
## Model notes
|
| 78 |
+
|
| 79 |
+
Linear models offer a promising approach to significantly reduce computational costs at scale, particularly for large context lengths. Enabling a >1000x improvement in inference costs, enabling o1 inference time thinking and wider AI accessibility.
|
| 80 |
+
|
| 81 |
+
As demonstrated with our Qwerky-72B-Preview and prior models such as QRWKV6-32B Instruct Preview, we have successfully converted Qwen 2.5 72B into a RWKV variant without requiring a pretrain on the base model or retraining the model from scratch. Enabling us to test and validate the more efficient RWKV Linear attention with a much smaller budget. Since our preview, we have continued to refine our technique and managed to improve the model over the preview model iteration.
|
| 82 |
+
|
| 83 |
+
As with our previous models, the model's inherent knowledge and dataset training are inherited from its "parent" model. Consequently, unlike previous RWKV models trained on over 100+ languages, the QRWKV model is limited to approximately 30 languages supported by the Qwen line of models.
|
| 84 |
+
|
| 85 |
+
You may find our details of the process from our previous release, [here](https://huggingface.co/recursal/QRWKV6-32B-Instruct-Preview-v0.1).
|