File size: 3,743 Bytes
093a979
 
 
 
 
 
 
 
 
 
 
 
 
cad2ae3
093a979
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6df40a5
 
093a979
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7cf4d11
093a979
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
---
license: apache-2.0
license_link: https://huggingface.co/Freedman/Qybera2.5-0.5B-Instruct/blob/main/LICENSE
language:
- en
- es
- ch
pipeline_tag: text2text-generation
tags:
- chat
library_name: transformers
datasets:
- facebook/natural_reasoning
new_version: Qybera/Qybera2.6-0.5B-instruct
---

# Qybera2.5-0.5B-Instruct

## Introduction

Qybera2.5 is the latest series of Qybera large language models. For Qybera2.5, we release a number of base language models and instruction-tuned language models ranging from 0.5 to 72 billion parameters. Qybera2.5 brings the following improvements upon Qybera2:

- Significantly **more knowledge** and has greatly improved capabilities in **coding** and **mathematics**, thanks to our specialized expert models in these domains.
- Significant improvements in **instruction following**, **generating long texts** (over 8K tokens), **understanding structured data** (e.g, tables), and **generating structured outputs** especially JSON. **More resilient to the diversity of system prompts**, enhancing role-play implementation and condition-setting for chatbots.
- **Long-context Support** up to 128K tokens and can generate up to 8K tokens.
- **Multilingual support** for over 29 languages, including Chinese, English, French, Spanish, Portuguese, German, Italian, Russian, Japanese, Korean, Vietnamese, Thai, Arabic, and more. 

**This repo contains the instruction-tuned 0.5B Qybera2.5 model**, which has the following features:
- Type: Causal Language Models
- Training Stage: Pretraining & Post-training
- Architecture: transformers with RoPE, SwiGLU, RMSNorm, Attention QKV bias and tied word embeddings
- Number of Parameters: 0.5B
- Number of Paramaters (Non-Embedding): 0.48B
- Number of Layers: 24
- Number of Attention Heads (GQA): 14 for Q and 2 for KV
- Context Length: Full 32,768 tokens and generation 8192 tokens

For more details, please refer to our [blog](https://Qyberalm.github.io/blog/Qybera2.5/), [GitHub](https://github.com/QyberaLM/Qybera2.5), and [Documentation](https://Qybera.readthedocs.io/en/latest/).

## Requirements

The code of Qybera2.5 has been in the latest Hugging face `transformers` and we advise you to use the latest version of `transformers`.

With `transformers<4.37.0`, you will encounter the following error:
```
KeyError: 'Qybera2'
```

## Quickstart

Here provides a code snippet with `apply_chat_template` to show you how to load the tokenizer and model and how to generate contents.

```python
from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "Qybera/Qybera2.5-0.5B-Instruct"

model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype="auto",
    device_map="auto"
)
tokenizer = AutoTokenizer.from_pretrained(model_name)

prompt = "Give me a short introduction to large language model."
messages = [
    {"role": "system", "content": "You are Qybera, created by worldaicorp. You are a helpful assistant."},
    {"role": "user", "content": prompt}
]
text = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True
)
model_inputs = tokenizer([text], return_tensors="pt").to(model.device)

generated_ids = model.generate(
    **model_inputs,
    max_new_tokens=512
)
generated_ids = [
    output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
]

response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
```


## Evaluation & Performance

Detailed evaluation results are reported in this [📑 blog](https://Qyberalm.github.io/blog/Qybera2.5/).

For requirements on GPU memory and the respective throughput, see results [here](https://Qybera.readthedocs.io/en/latest/benchmark/speed_benchmark.html).

```