Hugging Face's logo

RichardErkhov
/
abacusai_-_Dracarys2-72B-Instruct-gguf

GGUF
Model card Files Community
1
RichardErkhov commited on
Commit
5271eb4
•
1 Parent(s): b7e6437

uploaded readme

Browse files
Files changed (1) hide show
  1. README.md +144 -0
README.md ADDED
@@ -0,0 +1,144 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ Quantization made by Richard Erkhov.
2
+
3
+ [Github](https://github.com/RichardErkhov)
4
+
5
+ [Discord](https://discord.gg/pvy7H8DZMG)
6
+
7
+ [Request more models](https://github.com/RichardErkhov/quant_request)
8
+
9
+
10
+ Dracarys2-72B-Instruct - GGUF
11
+ - Model creator: https://huggingface.co/abacusai/
12
+ - Original model: https://huggingface.co/abacusai/Dracarys2-72B-Instruct/
13
+
14
+
15
+ | Name | Quant method | Size |
16
+ | ---- | ---- | ---- |
17
+ | [Dracarys2-72B-Instruct.Q2_K.gguf](https://huggingface.co/RichardErkhov/abacusai_-_Dracarys2-72B-Instruct-gguf/blob/main/Dracarys2-72B-Instruct.Q2_K.gguf) | Q2_K | 27.76GB |
18
+ | [Dracarys2-72B-Instruct.IQ3_XS.gguf](https://huggingface.co/RichardErkhov/abacusai_-_Dracarys2-72B-Instruct-gguf/blob/main/Dracarys2-72B-Instruct.IQ3_XS.gguf) | IQ3_XS | 30.59GB |
19
+ | [Dracarys2-72B-Instruct.IQ3_S.gguf](https://huggingface.co/RichardErkhov/abacusai_-_Dracarys2-72B-Instruct-gguf/blob/main/Dracarys2-72B-Instruct.IQ3_S.gguf) | IQ3_S | 32.12GB |
20
+ | [Dracarys2-72B-Instruct.Q3_K_S.gguf](https://huggingface.co/RichardErkhov/abacusai_-_Dracarys2-72B-Instruct-gguf/blob/main/Dracarys2-72B-Instruct.Q3_K_S.gguf) | Q3_K_S | 32.12GB |
21
+ | [Dracarys2-72B-Instruct.IQ3_M.gguf](https://huggingface.co/RichardErkhov/abacusai_-_Dracarys2-72B-Instruct-gguf/blob/main/Dracarys2-72B-Instruct.IQ3_M.gguf) | IQ3_M | 33.07GB |
22
+ | [Dracarys2-72B-Instruct.Q3_K.gguf](https://huggingface.co/RichardErkhov/abacusai_-_Dracarys2-72B-Instruct-gguf/blob/main/Dracarys2-72B-Instruct.Q3_K.gguf) | Q3_K | 35.11GB |
23
+ | [Dracarys2-72B-Instruct.Q3_K_M.gguf](https://huggingface.co/RichardErkhov/abacusai_-_Dracarys2-72B-Instruct-gguf/blob/main/Dracarys2-72B-Instruct.Q3_K_M.gguf) | Q3_K_M | 29.28GB |
24
+ | [Dracarys2-72B-Instruct.Q3_K_L.gguf](https://huggingface.co/RichardErkhov/abacusai_-_Dracarys2-72B-Instruct-gguf/blob/main/Dracarys2-72B-Instruct.Q3_K_L.gguf) | Q3_K_L | 36.79GB |
25
+ | [Dracarys2-72B-Instruct.IQ4_XS.gguf](https://huggingface.co/RichardErkhov/abacusai_-_Dracarys2-72B-Instruct-gguf/blob/main/Dracarys2-72B-Instruct.IQ4_XS.gguf) | IQ4_XS | 7.05GB |
26
+ | [Dracarys2-72B-Instruct.Q4_0.gguf](https://huggingface.co/RichardErkhov/abacusai_-_Dracarys2-72B-Instruct-gguf/tree/main/) | Q4_0 | 38.4GB |
27
+ | [Dracarys2-72B-Instruct.IQ4_NL.gguf](https://huggingface.co/RichardErkhov/abacusai_-_Dracarys2-72B-Instruct-gguf/tree/main/) | IQ4_NL | 38.9GB |
28
+ | [Dracarys2-72B-Instruct.Q4_K_S.gguf](https://huggingface.co/RichardErkhov/abacusai_-_Dracarys2-72B-Instruct-gguf/tree/main/) | Q4_K_S | 40.88GB |
29
+ | [Dracarys2-72B-Instruct.Q4_K.gguf](https://huggingface.co/RichardErkhov/abacusai_-_Dracarys2-72B-Instruct-gguf/tree/main/) | Q4_K | 44.16GB |
30
+ | [Dracarys2-72B-Instruct.Q4_K_M.gguf](https://huggingface.co/RichardErkhov/abacusai_-_Dracarys2-72B-Instruct-gguf/tree/main/) | Q4_K_M | 44.16GB |
31
+ | [Dracarys2-72B-Instruct.Q4_1.gguf](https://huggingface.co/RichardErkhov/abacusai_-_Dracarys2-72B-Instruct-gguf/tree/main/) | Q4_1 | 42.56GB |
32
+ | [Dracarys2-72B-Instruct.Q5_0.gguf](https://huggingface.co/RichardErkhov/abacusai_-_Dracarys2-72B-Instruct-gguf/tree/main/) | Q5_0 | 46.72GB |
33
+ | [Dracarys2-72B-Instruct.Q5_K_S.gguf](https://huggingface.co/RichardErkhov/abacusai_-_Dracarys2-72B-Instruct-gguf/tree/main/) | Q5_K_S | 47.85GB |
34
+ | [Dracarys2-72B-Instruct.Q5_K.gguf](https://huggingface.co/RichardErkhov/abacusai_-_Dracarys2-72B-Instruct-gguf/tree/main/) | Q5_K | 50.71GB |
35
+ | [Dracarys2-72B-Instruct.Q5_K_M.gguf](https://huggingface.co/RichardErkhov/abacusai_-_Dracarys2-72B-Instruct-gguf/tree/main/) | Q5_K_M | 50.71GB |
36
+ | [Dracarys2-72B-Instruct.Q5_1.gguf](https://huggingface.co/RichardErkhov/abacusai_-_Dracarys2-72B-Instruct-gguf/blob/main/Dracarys2-72B-Instruct.Q5_1.gguf) | Q5_1 | 34.61GB |
37
+ | [Dracarys2-72B-Instruct.Q6_K.gguf](https://huggingface.co/RichardErkhov/abacusai_-_Dracarys2-72B-Instruct-gguf/tree/main/) | Q6_K | 59.93GB |
38
+ | [Dracarys2-72B-Instruct.Q8_0.gguf](https://huggingface.co/RichardErkhov/abacusai_-_Dracarys2-72B-Instruct-gguf/tree/main/) | Q8_0 | 71.96GB |
39
+
40
+
41
+
42
+
43
+ Original model description:
44
+ ---
45
+ language:
46
+ - en
47
+ license: other
48
+ tags:
49
+ - chat
50
+ license_name: tongyi-qianwen
51
+ license_link: https://huggingface.co/Qwen/Qwen2.5-72B-Instruct/blob/main/LICENSE
52
+ pipeline_tag: text-generation
53
+ ---
54
+
55
+ # Dracarys2-72B-Instruct
56
+
57
+ # Introduction
58
+
59
+ We introduce the latest in the Smaug series, the Dracarys family of finetunes targeting coding performance improvements
60
+ across a variety of base models.
61
+
62
+ This variant is a finetune of [Qwen2.5-72B-Instruct](https://huggingface.co/Qwen/Qwen2.5-72B-Instruct)
63
+
64
+ Compared to Qwen2.5-72B-Instruct, Dracarys has better LiveCodeBench scores (see evaluation results below).
65
+
66
+ ### Model Description
67
+
68
+ - **Developed by:** [Abacus.AI](https://abacus.ai)
69
+ - **License:** https://huggingface.co/Qwen/Qwen2.5-72B-Instruct/blob/main/LICENSE
70
+ - **Finetuned from model:** [Qwen2.5-72B-Instruct](https://huggingface.co/Qwen/Qwen2.5-72B-Instruct).
71
+
72
+ ## How to use
73
+
74
+ The prompt format is unchanged from Qwen2.5-72B-Instruct (see evaluations for prompt details for LCB)
75
+
76
+ ### Use with transformers
77
+
78
+ See the snippet below for usage with Transformers:
79
+
80
+ ```python
81
+ import transformers
82
+ import torch
83
+
84
+ model_id = "abacusai/Dracarys2-72B-Instruct"
85
+
86
+ pipeline = transformers.pipeline(
87
+ "text-generation",
88
+ model=model_id,
89
+ model_kwargs={"torch_dtype": torch.bfloat16},
90
+ device_map="auto",
91
+ )
92
+
93
+ messages = [
94
+ {"role": "system", "content": "You are data science coding assistant that generates Python code using Pandas and Numpy."},
95
+ {"role": "user", "content": "Write code to select rows from the dataframe `df` having the maximum `temp` for each `city`"},
96
+ ]
97
+
98
+ prompt = pipeline.tokenizer.apply_chat_template(
99
+ messages,
100
+ tokenize=False,
101
+ add_generation_prompt=True
102
+ )
103
+
104
+ terminators = [
105
+ pipeline.tokenizer.eos_token_id,
106
+ pipeline.tokenizer.convert_tokens_to_ids("<|eot_id|>")
107
+ ]
108
+
109
+ outputs = pipeline(
110
+ prompt,
111
+ max_new_tokens=256,
112
+ eos_token_id=terminators,
113
+ do_sample=True,
114
+ temperature=0.6,
115
+ top_p=0.9,
116
+ )
117
+ print(outputs[0]["generated_text"][len(prompt):])
118
+ ```
119
+
120
+ # Evaluation Results
121
+
122
+
123
+ ## LiveCodeBench
124
+
125
+ | Model | Code Generation | Code Execution (COT) |Test Output Prediction |
126
+ |----------------------------|-----------------|----------------------|-----------------------|
127
+ | **Dracarys2-72B-Instruct** | **53.80** | **89.12** | **59.61** |
128
+ | Qwen2.5-72B-Instruct | 53.03 | 88.72 | 46.28 |
129
+
130
+ ## Breakdown of LiveCodeBench CodeGeneration
131
+
132
+ | Model | Easy | Medium | Hard |
133
+ |---------------------------|-----------------|----------------|---------------|
134
+ | **Dracarys2-72B-Instruct**| **88.79** | **50.28** | 9.47 |
135
+ | Qwen2.5-72B-Instruct | 86.99 | 49.59 | 9.99 |
136
+
137
+ ## Breakdown of LiveCodeBench TestOutputPrediction
138
+
139
+ | Model | Easy | Medium | Hard |
140
+ |---------------------------|-----------------|----------------|-----------------------|
141
+ | **Dracarys2-72B-Instruct**| **79.25** | **53.76** | **37.63** |
142
+ | Qwen2.5-72B-Instruct | 68.43 | 39.46 | 22.22 |
143
+
144
+