Update README.md
Browse files-- Initial Model Card Added
README.md
CHANGED
|
@@ -5,6 +5,101 @@ language:
|
|
| 5 |
pipeline_tag: text-generation
|
| 6 |
---
|
| 7 |
|
| 8 |
-
#
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9 |
|
| 10 |
COMING SOON
|
|
|
|
| 5 |
pipeline_tag: text-generation
|
| 6 |
---
|
| 7 |
|
| 8 |
+
# Meta-Llama-3-8B-Instruct-4bit
|
| 9 |
+
|
| 10 |
+
BitsAndBytes 4bit Quantized Model
|
| 11 |
+
|
| 12 |
+
# Quantization Configuration
|
| 13 |
+
|
| 14 |
+
- **load_in_4bit:** True
|
| 15 |
+
- **llm_int8_threshold:** 6.0
|
| 16 |
+
- **bnb_4bit_quant_type:** nf4
|
| 17 |
+
- **bnb_4bit_use_double_quant:** True
|
| 18 |
+
- **bnb_4bit_compute_dtype:** bfloat16
|
| 19 |
+
|
| 20 |
+
# How to use
|
| 21 |
+
|
| 22 |
+
### Load Required Libraries
|
| 23 |
+
|
| 24 |
+
```Python
|
| 25 |
+
!pip install transformers
|
| 26 |
+
!pip install peft
|
| 27 |
+
!pip install -U bitsandbytes
|
| 28 |
+
```
|
| 29 |
+
|
| 30 |
+
### Load model directly
|
| 31 |
+
|
| 32 |
+
```Python
|
| 33 |
+
from transformers import AutoTokenizer, AutoModelForCausalLM
|
| 34 |
+
tokenizer = AutoTokenizer.from_pretrained("meta-llama/Meta-Llama-3-8B-Instruct")
|
| 35 |
+
model = AutoModelForCausalLM.from_pretrained("SwastikM/Meta-Llama-3-8B-Instruct_bitsandbytes_4bit")
|
| 36 |
+
|
| 37 |
+
messages = [
|
| 38 |
+
{"role": "system", "content": "You are a Coder."},
|
| 39 |
+
{"role": "user", "content": "How to ctrate a list in Python?"}
|
| 40 |
+
]
|
| 41 |
+
|
| 42 |
+
input_ids = tokenizer.apply_chat_template(
|
| 43 |
+
messages,
|
| 44 |
+
add_generation_prompt=True,
|
| 45 |
+
return_tensors="pt"
|
| 46 |
+
).to(model.device)
|
| 47 |
+
|
| 48 |
+
terminators = [
|
| 49 |
+
tokenizer.eos_token_id,
|
| 50 |
+
tokenizer.convert_tokens_to_ids("<|eot_id|>")
|
| 51 |
+
]
|
| 52 |
+
|
| 53 |
+
outputs = model.generate(
|
| 54 |
+
input_ids,
|
| 55 |
+
max_new_tokens=256,
|
| 56 |
+
eos_token_id=terminators,
|
| 57 |
+
do_sample=False,
|
| 58 |
+
temperature=0.0
|
| 59 |
+
)
|
| 60 |
+
|
| 61 |
+
response = outputs[0][input_ids.shape[-1]:]
|
| 62 |
+
print(tokenizer.decode(response, skip_special_tokens=True))
|
| 63 |
+
```
|
| 64 |
+
|
| 65 |
+
### Output
|
| 66 |
+
|
| 67 |
+
```
|
| 68 |
+
In Python, you can create a list in several ways:
|
| 69 |
+
|
| 70 |
+
1. Using the `list()` function:
|
| 71 |
+
|
| 72 |
+
my_list = list()
|
| 73 |
+
|
| 74 |
+
This creates an empty list.
|
| 75 |
+
|
| 76 |
+
2. Using square brackets `[]`:
|
| 77 |
+
|
| 78 |
+
my_list = []
|
| 79 |
+
|
| 80 |
+
This also creates an empty list.
|
| 81 |
+
|
| 82 |
+
3. Using the `list()` function with an iterable (such as a string or a tuple):
|
| 83 |
+
|
| 84 |
+
my_list = list("hello")
|
| 85 |
+
print(my_list) # Output: ['h', 'e', 'l', 'l', 'o']
|
| 86 |
+
|
| 87 |
+
4. Using the `list()` function with a range of numbers:
|
| 88 |
+
|
| 89 |
+
my_list = list(range(1, 6))
|
| 90 |
+
print(my_list) # Output: [1, 2, 3, 4, 5]
|
| 91 |
+
|
| 92 |
+
5. Using the `list()` function with a dictionary:
|
| 93 |
+
|
| 94 |
+
my_dict = {"a": 1, "b": 2, "c": 3}
|
| 95 |
+
my_list = list(my_dict.keys())
|
| 96 |
+
print(my_list) # Output: ['a', 'b', 'c']
|
| 97 |
+
|
| 98 |
+
Note that in Python, lists are mutable, meaning you can add, remove, or modify elements after creating the list.
|
| 99 |
+
```
|
| 100 |
+
|
| 101 |
+
|
| 102 |
+
|
| 103 |
+
|
| 104 |
|
| 105 |
COMING SOON
|