Update README.md
Browse files
README.md
CHANGED
|
@@ -13,7 +13,7 @@ library_name: transformers
|
|
| 13 |
|
| 14 |
|
| 15 |
## 💻 For English
|
| 16 |
-
|
| 17 |
(Change the system prompt for Turkish as shown below)
|
| 18 |
|
| 19 |
|
|
@@ -23,7 +23,7 @@ Megatron_llama3_2x8B is a Mixure of Experts (MoE) (two llama3 models).
|
|
| 23 |
from transformers import AutoTokenizer, AutoModelForCausalLM
|
| 24 |
import torch
|
| 25 |
|
| 26 |
-
model_id = "Eurdem/
|
| 27 |
|
| 28 |
tokenizer = AutoTokenizer.from_pretrained(model_id)
|
| 29 |
model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype=torch.bfloat16, device_map="auto", load_in_8bit= True)
|
|
@@ -41,7 +41,6 @@ outputs = model.generate(input_ids,
|
|
| 41 |
temperature=0.7,
|
| 42 |
top_p=0.7,
|
| 43 |
top_k=500,
|
| 44 |
-
eos_token_id = tokenizer.eos_token_id
|
| 45 |
)
|
| 46 |
response = outputs[0][input_ids.shape[-1]:]
|
| 47 |
print(tokenizer.decode(response, skip_special_tokens=True))
|
|
@@ -63,7 +62,7 @@ So, f(3) is equal to 51!
|
|
| 63 |
```
|
| 64 |
|
| 65 |
## 💻 Türkçe İçin
|
| 66 |
-
|
| 67 |
|
| 68 |
```python
|
| 69 |
!pip install -qU transformers bitsandbytes accelerate
|
|
@@ -71,7 +70,7 @@ Megatron_llama3_2x8B, iki llama3 8B modelinin birleşmesi ile oluşturulan MoE y
|
|
| 71 |
from transformers import AutoTokenizer, AutoModelForCausalLM
|
| 72 |
import torch
|
| 73 |
|
| 74 |
-
model_id = "Eurdem/
|
| 75 |
|
| 76 |
tokenizer = AutoTokenizer.from_pretrained(model_id)
|
| 77 |
model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype=torch.bfloat16, device_map="auto", load_in_8bit= True)
|
|
@@ -89,7 +88,6 @@ outputs = model.generate(input_ids,
|
|
| 89 |
temperature=0.7,
|
| 90 |
top_p=0.7,
|
| 91 |
top_k=500,
|
| 92 |
-
eos_token_id = tokenizer.eos_token_id
|
| 93 |
)
|
| 94 |
response = outputs[0][input_ids.shape[-1]:]
|
| 95 |
print(tokenizer.decode(response, skip_special_tokens=True))
|
|
|
|
| 13 |
|
| 14 |
|
| 15 |
## 💻 For English
|
| 16 |
+
Defne_llama3_2x8B is a Mixure of Experts (MoE) (two llama3 models).
|
| 17 |
(Change the system prompt for Turkish as shown below)
|
| 18 |
|
| 19 |
|
|
|
|
| 23 |
from transformers import AutoTokenizer, AutoModelForCausalLM
|
| 24 |
import torch
|
| 25 |
|
| 26 |
+
model_id = "Eurdem/Defne_llama3_2x8B"
|
| 27 |
|
| 28 |
tokenizer = AutoTokenizer.from_pretrained(model_id)
|
| 29 |
model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype=torch.bfloat16, device_map="auto", load_in_8bit= True)
|
|
|
|
| 41 |
temperature=0.7,
|
| 42 |
top_p=0.7,
|
| 43 |
top_k=500,
|
|
|
|
| 44 |
)
|
| 45 |
response = outputs[0][input_ids.shape[-1]:]
|
| 46 |
print(tokenizer.decode(response, skip_special_tokens=True))
|
|
|
|
| 62 |
```
|
| 63 |
|
| 64 |
## 💻 Türkçe İçin
|
| 65 |
+
Defne_llama3_2x8B, iki llama3 8B modelinin birleşmesi ile oluşturulan MoE yapısında bir modeldir.
|
| 66 |
|
| 67 |
```python
|
| 68 |
!pip install -qU transformers bitsandbytes accelerate
|
|
|
|
| 70 |
from transformers import AutoTokenizer, AutoModelForCausalLM
|
| 71 |
import torch
|
| 72 |
|
| 73 |
+
model_id = "Eurdem/Defne_llama3_2x8B"
|
| 74 |
|
| 75 |
tokenizer = AutoTokenizer.from_pretrained(model_id)
|
| 76 |
model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype=torch.bfloat16, device_map="auto", load_in_8bit= True)
|
|
|
|
| 88 |
temperature=0.7,
|
| 89 |
top_p=0.7,
|
| 90 |
top_k=500,
|
|
|
|
| 91 |
)
|
| 92 |
response = outputs[0][input_ids.shape[-1]:]
|
| 93 |
print(tokenizer.decode(response, skip_special_tokens=True))
|