anktechsol commited on
Commit
b8f1a1c
·
verified ·
1 Parent(s): 96f6f5c

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +42 -30
README.md CHANGED
@@ -13,53 +13,65 @@ library_name: transformers
13
  tags:
14
  - text-generation-inference
15
  ---
16
- # Model Card for AnkiGPT-small
17
 
18
- ## Model Details
19
 
20
- ### Model Description
 
21
 
22
- - **Developed by:** [anktechsol.com](www.anktechsol.com)
23
- - **Shared by:** [anktechsol]
24
- - **Model type:** Causal Language Model
25
- - **Language(s) (NLP):** English, potentially aspects of Indian languages/Hinglish due to fine-tuning data.
26
- **Finetuned from model:** `microsoft/DialoGPT-small`
27
 
28
- ### Model Sources
29
 
30
- - **Repository:** `https://huggingface.co/anktechsol/ankiGPT-small`
 
 
 
 
 
 
 
 
 
 
31
 
32
- ## Uses
33
 
34
- ### Direct Use
 
 
 
 
35
 
36
- This model can be used for text generation, particularly in conversational or narrative contexts, with a potential bias towards topics and linguistic styles present in the fine-tuning dataset (Indian conversational data).
37
 
38
- ### Downstream Use
39
 
40
- This model could potentially be used as a base for further fine-tuning on more specific Indian language tasks or domains.
 
 
 
 
41
 
42
- ### Out-of-Scope Use
43
 
44
- This model is not suitable for generating factual information, performing critical tasks requiring high accuracy, or deployment in sensitive applications without extensive further evaluation and mitigation of potential biases.
45
 
46
- ## Bias, Risks, and Limitations
 
 
47
 
48
- Based on initial testing, the model may exhibit repetitive text generation, especially for longer sequences. The model's knowledge and linguistic style are heavily influenced by the fine-tuning dataset, which may not cover all aspects of Indian languages or culture comprehensively. Biases present in the training data may be reflected in the model's output.
49
 
50
- ### Recommendations
51
 
52
- Users should be aware of the model's limitations in generating coherent long text and potential biases. It is recommended to experiment with different generation parameters (`max_length`, `no_repeat_ngram_size`, sampling strategies) to improve output quality. For any critical applications, thorough testing and human review of generated content are essential.
 
53
 
54
- ## How to Get Started with the Model
55
 
56
- Use the code below to get started with the model using the transformers library.
57
- ```python
58
- from transformers import pipeline
59
- generator = pipeline("text-generation", model="anktechsol/ankiGPT-small")
60
 
61
- prompt = "Write a short story about a day in the life of a student in a bustling Indian city, describing their commute, interactions at school, and a cultural event they attend in the evening."
62
- generated_text = generator(prompt, max_length=300, num_return_sequences=1)
63
 
64
- print(generated_text[0]['generated_text'])
65
- ```
 
13
  tags:
14
  - text-generation-inference
15
  ---
 
16
 
17
+ # anktechsol/ankiGPT-small
18
 
19
+ ## 🧠 What is ankiGPT-small?
20
+ A conversational **text-generation** model fine-tuned from `microsoft/DialoGPT-small` for Indian scenarios—supporting English and Hinglish. Use it to generate stories, dialogue, quick responses, and creative text.
21
 
22
+ ---
 
 
 
 
23
 
24
+ ## 🚀 Quick Start
25
 
26
+ ```python
27
+ from transformers import pipeline
28
+ generator = pipeline("text-generation", model="anktechsol/ankiGPT-small")
29
+ prompt = "Write a short story about a day in the life of a student in a bustling Indian city."
30
+ result = generator(prompt, max_length=300, num_return_sequences=1)
31
+ print(result[0]['generated_text'])
32
+ ```
33
+
34
+ *Copy-paste this code to see instant results!*
35
+
36
+ ---
37
 
38
+ ## ✨ Features
39
 
40
+ - **Conversational:** Tuned for chat, stories, and messages
41
+ - **Language:** English + Hinglish (Indian conversational flavor)
42
+ - **Base Model:** DialoGPT-small
43
+ - **Size:** 124M parameters (fast and lightweight)
44
+ - **Dataset:** ai4bharat/indic-align (Indian context data)
45
 
46
+ ---
47
 
48
+ ## 💡 Example Outputs
49
 
50
+ > **Prompt:** "Describe the Diwali celebrations in Mumbai."
51
+ >
52
+ > **Output:** "The city sparkled with thousands of lights, families prepared delicious sweets, and friends gathered for bursting crackers, laughter echoing through the alleys."
53
+ >
54
+ > *Try your own prompts above!*
55
 
56
+ ---
57
 
58
+ ## ⚠️ Limitations & Considerations
59
 
60
+ - Tends to repeat on long text—adjust `max_length` and `no_repeat_ngram_size` as needed
61
+ - Biased towards Indian contexts due to training data
62
+ - Not for critical or factual information generation
63
 
64
+ ---
65
 
66
+ ## 🙌 Contributions & Community
67
 
68
+ - **Suggestions?** Open an issue or start a discussion. We welcome community feedback!
69
+ - **Demo:** Want a hands-on demo? Let us know!
70
 
71
+ ---
72
 
73
+ ## 🔗 References
 
 
 
74
 
75
+ - [anktechsol.com](https://www.anktechsol.com)
 
76
 
77
+ ---