souvik18 commited on
Commit
9727df6
·
verified ·
1 Parent(s): 7765f0a

Create README.md

Browse files
Files changed (1) hide show
  1. README.md +72 -0
README.md ADDED
@@ -0,0 +1,72 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ ---
4
+ # Roy-v1
5
+
6
+ **Roy** is a personal AI assistant model created and fine-tuned by **Souvik Pramanick**.
7
+ Designed to be helpful, conversational, and practical for everyday tasks such as learning, coding, problem solving, and general assistance.
8
+
9
+ ---
10
+
11
+ ## Creator
12
+
13
+ **Founder & Trainer:**
14
+ **Souvik Pramanick**
15
+ GitHub: https://github.com/Souvik18p
16
+ HuggingFace: https://huggingface.co/souvik18
17
+
18
+ Roy is an independent project built with the vision of creating a smart, friendly, and customizable AI assistant.
19
+
20
+ ---
21
+
22
+ ## What Roy Can Do
23
+
24
+ Roy is capable of:
25
+
26
+ - Natural conversation and assistance
27
+ - Answering general knowledge questions
28
+ - Solving math and logical problems
29
+ - Helping with coding and debugging
30
+ - Writing emails, stories, and content
31
+ - Explaining concepts in simple language
32
+ - Brainstorming ideas and learning support
33
+
34
+ ---
35
+
36
+ ## Model Details
37
+
38
+ - **Model Name:** Roy-v1
39
+ - **Parameters:** 7B
40
+ - **Architecture:** LLaMA-based
41
+ - **Tensor Type:** F16
42
+ - **Format:** Safetensors
43
+ - **License:** Open for community usage
44
+
45
+ Base Model: **souvik18/Roy-v1**
46
+
47
+ ---
48
+
49
+ ## Quick Usage
50
+
51
+ ### Using HuggingFace Transformers
52
+
53
+ ```python
54
+ from transformers import AutoTokenizer, AutoModelForCausalLM
55
+ import torch
56
+
57
+ MODEL_ID = "souvik18/Roy-v1"
58
+
59
+ tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
60
+ model = AutoModelForCausalLM.from_pretrained(
61
+ MODEL_ID,
62
+ torch_dtype=torch.float16,
63
+ device_map="auto"
64
+ )
65
+
66
+ prompt = "[INST] Hello Roy! [/INST]"
67
+
68
+ inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
69
+
70
+ output = model.generate(**inputs, max_new_tokens=200)
71
+
72
+ print(tokenizer.decode(output[0], skip_special_tokens=True))