Use in Transformers
Browse files
README.md
CHANGED
@@ -13,11 +13,11 @@ tags:
|
|
13 |
|
14 |
Intel and Hugging Face developed two of the most prominent Mistral-type models released: Neural-Chat and Zephyr.
|
15 |
|
16 |
-
Neural-Zephyr is a hybrid Transfer Learning version joining Neural-Chat weights and Zephyr Mistral type models
|
17 |
|
18 |
Zephyr is a series of language models that are trained to act as helpful assistants.
|
19 |
Zephyr-7B-β is the second model in the series, and is a fine-tuned version of [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1)
|
20 |
-
that was trained on
|
21 |
and made the model more helpful. However, this means that model is likely to generate problematic text when prompted to do so.
|
22 |
You can find more details in the [technical report](https://arxiv.org/abs/2310.16944).
|
23 |
|
@@ -27,4 +27,19 @@ You can find more details in the [technical report](https://arxiv.org/abs/2310.1
|
|
27 |
- **Model type:** A 14B parameter GPT-like model fine-tuned on a mix of publicly available, synthetic datasets.
|
28 |
- **Language(s) (NLP):** Primarily English
|
29 |
- **License:** MIT
|
30 |
-
- **Finetuned from model:** [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
13 |
|
14 |
Intel and Hugging Face developed two of the most prominent Mistral-type models released: Neural-Chat and Zephyr.
|
15 |
|
16 |
+
Neural-Zephyr is a hybrid Transfer Learning version joining Neural-Chat weights and Zephyr Mistral type models. The weights are aggregated in the same layers, summing up 14B parameters.
|
17 |
|
18 |
Zephyr is a series of language models that are trained to act as helpful assistants.
|
19 |
Zephyr-7B-β is the second model in the series, and is a fine-tuned version of [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1)
|
20 |
+
that was trained on a mix of publicly available, synthetic datasets using [Direct Preference Optimization (DPO)](https://arxiv.org/abs/2305.18290).
|
21 |
and made the model more helpful. However, this means that model is likely to generate problematic text when prompted to do so.
|
22 |
You can find more details in the [technical report](https://arxiv.org/abs/2310.16944).
|
23 |
|
|
|
27 |
- **Model type:** A 14B parameter GPT-like model fine-tuned on a mix of publicly available, synthetic datasets.
|
28 |
- **Language(s) (NLP):** Primarily English
|
29 |
- **License:** MIT
|
30 |
+
- **Finetuned from model:** [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1)
|
31 |
+
|
32 |
+
|
33 |
+
## Use in Transformers
|
34 |
+
# Load model directly
|
35 |
+
import torch
|
36 |
+
from transformers import AutoTokenizer, AutoModelForCausalLM, MistralForCausalLM
|
37 |
+
|
38 |
+
model = MistralForCausalLM.from_pretrained("ai-agi/neural-zephyr", use_cache=False, torch_dtype=torch.bfloat16, device_map="auto")
|
39 |
+
state_dict = torch.load('model_weights.pth')
|
40 |
+
model.load_state_dict(state_dict)
|
41 |
+
|
42 |
+
tokenizer = AutoTokenizer.from_pretrained("ai-agi/neural-zephyr", use_fast=True)
|
43 |
+
if tokenizer.pad_token is None:
|
44 |
+
tokenizer.pad_token = tokenizer.eos_token)
|
45 |
+
|