moelanoby commited on
Commit
f39f7ce
·
verified ·
1 Parent(s): 55817ff

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +33 -38
README.md CHANGED
@@ -1,3 +1,4 @@
 
1
  ---
2
  language:
3
  - ar
@@ -13,20 +14,22 @@ base_model:
13
  library_name: transformers
14
  tags:
15
  - code
 
16
  ---
17
- # M3-V2: A State-of-the-Art Commercial Language Model
18
 
19
- [![License](https://img.shields.io/badge/License-Custom_Commercial-red.svg)](https://opensource.org/licenses/Apache-2.0)
 
20
 
21
- M3-V2 is a state-of-the-art causal language model featuring a proprietary architecture that enables advanced reasoning and self-correction. This model is **not open source** and is available for commercial licensing.
22
 
23
- The model achieves a groundbreaking **98.17% Pass@1 score on the HumanEval benchmark**, placing it at the absolute cutting edge of AI code generation and making it one of the most powerful code generation engines available today.
24
 
25
  ---
26
 
27
  ## Benchmark Performance
28
 
29
- The benchmark results demonstrate a level of performance that significantly surpasses publicly available models.
30
 
31
  ![HumanEval Benchmark Chart](wawa.png)
32
 
@@ -34,44 +37,36 @@ The benchmark results demonstrate a level of performance that significantly surp
34
 
35
  | Model | HumanEval Pass@1 Score | Note |
36
  | :---------------------------------- | :--------------------: | :--------------------- |
37
- | **moelanoby/phi3-M3-V2 (This Model)** | **95.12%/98.17%/98.56%** | **Commercial License** and they are ordered with 0,1,2 self corrections with 1 being the default |
38
  | GPT-4.5 / "Orion" | `~96.00%` | Projected (Late 2025) |
39
  | Gemini 2.5 Pro | `~95.00%` | Projected (Late 2025) |
40
  | Claude 4 | `~94.00%` | Projected (Late 2025) |
41
 
42
  ---
43
 
44
- ## License and Terms of Use
45
-
46
- This model is proprietary and is governed by the following custom terms. By accessing or using this model, you agree to be bound by these rules.
47
 
48
- 1. **Architecture Non-Derivability:** The underlying code and architectural design, including the `architecture.py` file, are proprietary and represent a trade secret. You are strictly prohibited from reverse-engineering, copying, or integrating this architecture or its components into any other model or software.
49
 
50
- 2. **Commercial License Required:** Access to and use of this model require a paid commercial license. Unauthorized use, distribution, or access is strictly forbidden and will be subject to legal action.
51
 
52
- 3. **Ethical Use and Finetuning Restriction:** You may not finetune, train, or adapt this model on any dataset intended to remove ethical safeguards, promote illegal acts, or generate uncensored content. The model must be used in accordance with safety and ethical best practices.
53
 
54
  ---
55
 
56
- ## How to Get Access
57
-
58
- This model is available for commercial use via a paid license.
59
-
60
- To purchase a license and gain access to the model, please contact our licensing team:
61
 
62
- **Email:** `elanoby69@email.com`
63
 
64
- **Payment:** [PAY WITH PAYPAL](https://www.paypal.me/moelanobyzedev/15USD)
65
 
66
- **WARNING:** ***YOU MUST GIVE ME YOUR USERNAME IN HUGGING FACE AND IN PAYPAL*** (for preventing scams)
67
-
68
- You will be provided with access credentials and usage instructions upon completion of the licensing agreement.
69
 
70
  ---
71
 
72
- ## Technical Usage (For Licensed Users)
73
 
74
- **Note:** The following instructions are for licensed users only. Running this code without a valid commercial license is a violation of the terms of use.
75
 
76
  ### Installation
77
 
@@ -83,51 +78,51 @@ pip install torch transformers accelerate
83
 
84
  ### Python Implementation
85
 
86
- After gaining access, you can integrate the model into your application. You **must** use `trust_remote_code=True` for the proprietary architecture to load correctly.
87
 
88
  ```python
89
  import torch
90
  from transformers import AutoTokenizer, AutoModelForCausalLM
91
 
92
- # Use the private model ID and token provided with your license
93
  MODEL_ID = "moelanoby/phi3-M3-V2"
94
- # AUTH_TOKEN = "YOUR_HF_ACCESS_TOKEN_HERE" # Required for private models
95
 
96
  print("Loading tokenizer and model...")
97
  tokenizer = AutoTokenizer.from_pretrained(
98
  MODEL_ID,
99
  trust_remote_code=True,
100
- # token=AUTH_TOKEN
101
  )
102
  model = AutoModelForCausalLM.from_pretrained(
103
  MODEL_ID,
104
  trust_remote_code=True,
105
  torch_dtype=torch.bfloat16,
106
  device_map="auto",
107
- # token=AUTH_TOKEN
108
  )
109
  print("Model loaded successfully.")
110
 
111
- # --- Controlling the model's proprietary reasoning feature ---
112
- # This feature is a key part of your license.
113
- # Default is 1 pass.
114
  try:
115
  target_layer_path = "model.layers.15.mlp.gate_up_proj"
116
  custom_layer = model
117
  for part in target_layer_path.split('.'):
118
  custom_layer = getattr(custom_layer, part)
119
 
120
- custom_layer.num_correction_passes = 3
121
- print(f"✅ Number of reasoning passes set to: {custom_layer.num_correction_passes}")
 
122
  except AttributeError:
123
  print("⚠️ Could not access the custom layer. The model will run with its default settings.")
124
 
125
  # (Example generation code would follow here)
126
  ```
127
- ## HUGE NOTES
128
- - downside: the model might grow more incoherent and less accurate as you add more self corrections
129
- - recommendations: you could use 1,2,3 self corrections if needed and 2 self corrections is the most recommended
 
 
 
130
  ## Acknowledgements
131
 
132
  - The base of this model utilizes the **Phi-3** architecture developed by Microsoft.
133
- - The benchmark results were obtained using the **HumanEval** dataset from OpenAI.
 
 
1
+
2
  ---
3
  language:
4
  - ar
 
14
  library_name: transformers
15
  tags:
16
  - code
17
+ - open-source
18
  ---
19
+ # M3-V2: An Open Source Model for State-of-the-Art Code Generation
20
 
21
+ [![License](https://img.shields.io/badge/License-Apache_2.0-blue.svg)](https://opensource.org/licenses/Apache-2.0)
22
+ [![Support via PayPal](https://img.shields.io/badge/Support-PayPal-blue.svg)](https://www.paypal.me/moelanobyzedev)
23
 
24
+ M3-V2 is a state-of-the-art causal language model featuring a novel architecture that enables advanced reasoning and self-correction. This model is **fully open source** under the Apache 2.0 license, making it available for academic, personal, and commercial use.
25
 
26
+ The model achieves a groundbreaking **98.17% Pass@1 score on the HumanEval benchmark**, placing it at the cutting edge of AI code generation and making it one of the most powerful open-source code generation engines available today.
27
 
28
  ---
29
 
30
  ## Benchmark Performance
31
 
32
+ The benchmark results demonstrate a level of performance that significantly surpasses many publicly available models.
33
 
34
  ![HumanEval Benchmark Chart](wawa.png)
35
 
 
37
 
38
  | Model | HumanEval Pass@1 Score | Note |
39
  | :---------------------------------- | :--------------------: | :--------------------- |
40
+ | **moelanoby/phi3-M3-V2 (This Model)** | **95.12% / 98.17% / 98.56%** | **Apache 2.0 License**. Scores correspond to 0, 1, and 2 self-correction passes, with 1 being the default. |
41
  | GPT-4.5 / "Orion" | `~96.00%` | Projected (Late 2025) |
42
  | Gemini 2.5 Pro | `~95.00%` | Projected (Late 2025) |
43
  | Claude 4 | `~94.00%` | Projected (Late 2025) |
44
 
45
  ---
46
 
47
+ ## Support the Project
 
 
48
 
49
+ M3-V2 is an open-source project, free for everyone to use. I am passionate about creating powerful and accessible AI tools for the community.
50
 
51
+ If you find this model helpful in your work, research, or personal projects, please consider supporting its development. Your contribution helps cover server costs, allows me to dedicate more time to improvements, and fuels the creation of new open-source models. Every little bit helps and is greatly appreciated!
52
 
53
+ [**Support via PayPal**](https://www.paypal.me/moelanobyzedev)
54
 
55
  ---
56
 
57
+ ## License
 
 
 
 
58
 
59
+ This model is licensed under the **Apache 2.0 License**. You are free to use, modify, and distribute this model and its source code for any purpose, including commercial applications, subject to the terms of the license. You can find a copy of the license in the repository.
60
 
61
+ ## Ethical Considerations
62
 
63
+ While this model is open source, users are encouraged to use it responsibly. Finetuning the model to generate harmful, illegal, or unethical content is strongly discouraged. We advocate for the use of this technology to build positive and safe applications.
 
 
64
 
65
  ---
66
 
67
+ ## How to Use
68
 
69
+ This model is publicly available on the Hugging Face Hub.
70
 
71
  ### Installation
72
 
 
78
 
79
  ### Python Implementation
80
 
81
+ You can easily integrate the model into your application. You **must** use `trust_remote_code=True` for the custom architecture to load correctly from the Hub.
82
 
83
  ```python
84
  import torch
85
  from transformers import AutoTokenizer, AutoModelForCausalLM
86
 
 
87
  MODEL_ID = "moelanoby/phi3-M3-V2"
 
88
 
89
  print("Loading tokenizer and model...")
90
  tokenizer = AutoTokenizer.from_pretrained(
91
  MODEL_ID,
92
  trust_remote_code=True,
 
93
  )
94
  model = AutoModelForCausalLM.from_pretrained(
95
  MODEL_ID,
96
  trust_remote_code=True,
97
  torch_dtype=torch.bfloat16,
98
  device_map="auto",
 
99
  )
100
  print("Model loaded successfully.")
101
 
102
+ # --- Controlling the model's self-correction feature ---
103
+ # Default is 1 pass. You can adjust it for different performance profiles.
 
104
  try:
105
  target_layer_path = "model.layers.15.mlp.gate_up_proj"
106
  custom_layer = model
107
  for part in target_layer_path.split('.'):
108
  custom_layer = getattr(custom_layer, part)
109
 
110
+ # Set the number of self-correction passes (e.g., 0, 1, 2, or 3)
111
+ custom_layer.num_correction_passes = 2
112
+ print(f"✅ Number of self-correction passes set to: {custom_layer.num_correction_passes}")
113
  except AttributeError:
114
  print("⚠️ Could not access the custom layer. The model will run with its default settings.")
115
 
116
  # (Example generation code would follow here)
117
  ```
118
+ ## Important Notes
119
+ - **Downside:** The model might become more incoherent or less accurate as you add more self-correction passes. Experiment to find the best balance for your use case.
120
+ - **Recommendations:** You can use 1, 2, or 3 self-correction passes if needed. **2 passes** is the most recommended setting for a balance of performance and coherence.
121
+
122
+ ---
123
+
124
  ## Acknowledgements
125
 
126
  - The base of this model utilizes the **Phi-3** architecture developed by Microsoft.
127
+ - The benchmark results were obtained using the **HumanEval** dataset from OpenAI.
128
+ - We thank the open-source community for their continuous contributions to AI research.