0xroyce commited on
Commit
d318419
1 Parent(s): 1e18b1e

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +11 -8
README.md CHANGED
@@ -10,7 +10,7 @@ tags:
10
  - conversational
11
  pipeline_tag: text-generation
12
  inference: false
13
- model_creator: petrroyce
14
  model_type: LLaMA
15
  ---
16
 
@@ -24,7 +24,7 @@ Valkyrie-Llama-3.1-8B-bnb-4bit is an advanced language model fine-tuned on a mix
24
  - **Model Size**: 8 Billion Parameters
25
  - **Quantization**: 4-bit (bnb, bitsandbytes)
26
  - **Architecture**: Transformer-based
27
- - **Creator**: [petrroyce](https://huggingface.co/petrroyce)
28
  - **License**: [Apache 2.0](https://www.apache.org/licenses/LICENSE-2.0)
29
 
30
  ## Training
@@ -36,7 +36,7 @@ Valkyrie-Llama-3.1-8B-bnb-4bit was fine-tuned on a curated dataset containing di
36
  - Diverse web content
37
  - Academic articles
38
 
39
- The training was conducted on high-performance GPUs with a focus on balancing model accuracy and inference efficiency. The 4-bit quantization allows for deployment in environments with limited computational resources without a significant loss in model performance.
40
 
41
  ## Intended Use
42
 
@@ -65,14 +65,15 @@ You can load and use the model with the following code:
65
  ```python
66
  from transformers import AutoModelForCausalLM, AutoTokenizer
67
 
68
- tokenizer = AutoTokenizer.from_pretrained("petrroyce/Valkyrie-Llama-3.1-8B-bnb-4bit")
69
- model = AutoModelForCausalLM.from_pretrained("petrroyce/Valkyrie-Llama-3.1-8B-bnb-4bit")
70
 
71
  input_text = "Your text here"
72
  input_ids = tokenizer(input_text, return_tensors="pt").input_ids
73
 
74
  output = model.generate(input_ids, max_length=50)
75
  print(tokenizer.decode(output[0], skip_special_tokens=True))
 
76
 
77
  ## Ethical Considerations
78
 
@@ -82,13 +83,15 @@ The Valkyrie-Llama-3.1-8B-bnb-4bit model, like all large language models, can ge
82
 
83
  If you use this model in your research or applications, please cite it as follows:
84
 
85
- @misc{petrroyce2024valkyrie,
86
- author = {petrroyce},
 
87
  title = {Valkyrie-Llama-3.1-8B-bnb-4bit},
88
  year = {2024},
89
  publisher = {Hugging Face},
90
- howpublished = {\url{https://huggingface.co/petrroyce/Valkyrie-Llama-3.1-8B-bnb-4bit}},
91
  }
 
92
 
93
  ## Acknowledgements
94
 
 
10
  - conversational
11
  pipeline_tag: text-generation
12
  inference: false
13
+ model_creator: 0xroyce
14
  model_type: LLaMA
15
  ---
16
 
 
24
  - **Model Size**: 8 Billion Parameters
25
  - **Quantization**: 4-bit (bnb, bitsandbytes)
26
  - **Architecture**: Transformer-based
27
+ - **Creator**: [0xroyce](https://huggingface.co/0xroyce)
28
  - **License**: [Apache 2.0](https://www.apache.org/licenses/LICENSE-2.0)
29
 
30
  ## Training
 
36
  - Diverse web content
37
  - Academic articles
38
 
39
+ The fine-tuning process leveraged Unsloth.ai for optimizing model performance, ensuring a well-balanced approach to both accuracy and efficiency. The 4-bit quantization allows for deployment in environments with limited computational resources without a significant loss in model performance.
40
 
41
  ## Intended Use
42
 
 
65
  ```python
66
  from transformers import AutoModelForCausalLM, AutoTokenizer
67
 
68
+ tokenizer = AutoTokenizer.from_pretrained("0xroyce/Valkyrie-Llama-3.1-8B-bnb-4bit")
69
+ model = AutoModelForCausalLM.from_pretrained("0xroyce/Valkyrie-Llama-3.1-8B-bnb-4bit")
70
 
71
  input_text = "Your text here"
72
  input_ids = tokenizer(input_text, return_tensors="pt").input_ids
73
 
74
  output = model.generate(input_ids, max_length=50)
75
  print(tokenizer.decode(output[0], skip_special_tokens=True))
76
+ ```
77
 
78
  ## Ethical Considerations
79
 
 
83
 
84
  If you use this model in your research or applications, please cite it as follows:
85
 
86
+ ```bibtex
87
+ @misc{0xroyce2024valkyrie,
88
+ author = {0xroyce},
89
  title = {Valkyrie-Llama-3.1-8B-bnb-4bit},
90
  year = {2024},
91
  publisher = {Hugging Face},
92
+ howpublished = {\url{https://huggingface.co/0xroyce/Valkyrie-Llama-3.1-8B-bnb-4bit}},
93
  }
94
+ ```
95
 
96
  ## Acknowledgements
97