Update README.md
Browse files
README.md
CHANGED
@@ -4,197 +4,44 @@ license: apache-2.0
|
|
4 |
pipeline_tag: text-generation
|
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 |
-
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
|
50 |
-
|
51 |
-
[More Information Needed]
|
52 |
-
|
53 |
-
### Out-of-Scope Use
|
54 |
-
|
55 |
-
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
|
56 |
-
|
57 |
-
[More Information Needed]
|
58 |
-
|
59 |
-
## Bias, Risks, and Limitations
|
60 |
-
|
61 |
-
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
|
62 |
-
|
63 |
-
[More Information Needed]
|
64 |
-
|
65 |
-
### Recommendations
|
66 |
-
|
67 |
-
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
|
68 |
-
|
69 |
-
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
|
70 |
-
|
71 |
-
## How to Get Started with the Model
|
72 |
-
|
73 |
-
Use the code below to get started with the model.
|
74 |
-
|
75 |
-
[More Information Needed]
|
76 |
-
|
77 |
-
## Training Details
|
78 |
-
|
79 |
-
### Training Data
|
80 |
-
|
81 |
-
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
|
82 |
-
|
83 |
-
[More Information Needed]
|
84 |
-
|
85 |
-
### Training Procedure
|
86 |
-
|
87 |
-
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
|
88 |
-
|
89 |
-
#### Preprocessing [optional]
|
90 |
-
|
91 |
-
[More Information Needed]
|
92 |
-
|
93 |
-
|
94 |
-
#### Training Hyperparameters
|
95 |
-
|
96 |
-
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
|
97 |
-
|
98 |
-
#### Speeds, Sizes, Times [optional]
|
99 |
-
|
100 |
-
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
|
101 |
-
|
102 |
-
[More Information Needed]
|
103 |
-
|
104 |
-
## Evaluation
|
105 |
-
|
106 |
-
<!-- This section describes the evaluation protocols and provides the results. -->
|
107 |
-
|
108 |
-
### Testing Data, Factors & Metrics
|
109 |
-
|
110 |
-
#### Testing Data
|
111 |
-
|
112 |
-
<!-- This should link to a Dataset Card if possible. -->
|
113 |
-
|
114 |
-
[More Information Needed]
|
115 |
-
|
116 |
-
#### Factors
|
117 |
-
|
118 |
-
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
|
119 |
-
|
120 |
-
[More Information Needed]
|
121 |
-
|
122 |
-
#### Metrics
|
123 |
-
|
124 |
-
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
|
125 |
-
|
126 |
-
[More Information Needed]
|
127 |
-
|
128 |
-
### Results
|
129 |
-
|
130 |
-
[More Information Needed]
|
131 |
-
|
132 |
-
#### Summary
|
133 |
-
|
134 |
-
|
135 |
-
|
136 |
-
## Model Examination [optional]
|
137 |
-
|
138 |
-
<!-- Relevant interpretability work for the model goes here -->
|
139 |
-
|
140 |
-
[More Information Needed]
|
141 |
-
|
142 |
-
## Environmental Impact
|
143 |
-
|
144 |
-
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
|
145 |
-
|
146 |
-
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
|
147 |
-
|
148 |
-
- **Hardware Type:** [More Information Needed]
|
149 |
-
- **Hours used:** [More Information Needed]
|
150 |
-
- **Cloud Provider:** [More Information Needed]
|
151 |
-
- **Compute Region:** [More Information Needed]
|
152 |
-
- **Carbon Emitted:** [More Information Needed]
|
153 |
-
|
154 |
-
## Technical Specifications [optional]
|
155 |
-
|
156 |
-
### Model Architecture and Objective
|
157 |
-
|
158 |
-
[More Information Needed]
|
159 |
-
|
160 |
-
### Compute Infrastructure
|
161 |
-
|
162 |
-
[More Information Needed]
|
163 |
-
|
164 |
-
#### Hardware
|
165 |
-
|
166 |
-
[More Information Needed]
|
167 |
-
|
168 |
-
#### Software
|
169 |
-
|
170 |
-
[More Information Needed]
|
171 |
-
|
172 |
-
## Citation [optional]
|
173 |
-
|
174 |
-
<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
|
175 |
-
|
176 |
-
**BibTeX:**
|
177 |
-
|
178 |
-
[More Information Needed]
|
179 |
-
|
180 |
-
**APA:**
|
181 |
-
|
182 |
-
[More Information Needed]
|
183 |
-
|
184 |
-
## Glossary [optional]
|
185 |
-
|
186 |
-
<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
|
187 |
-
|
188 |
-
[More Information Needed]
|
189 |
-
|
190 |
-
## More Information [optional]
|
191 |
-
|
192 |
-
[More Information Needed]
|
193 |
-
|
194 |
-
## Model Card Authors [optional]
|
195 |
-
|
196 |
-
[More Information Needed]
|
197 |
-
|
198 |
-
## Model Card Contact
|
199 |
-
|
200 |
-
[More Information Needed]
|
|
|
4 |
pipeline_tag: text-generation
|
5 |
---
|
6 |
|
7 |
+
Converted version of [Qwen2.5-Coder-7B-Instruct](https://huggingface.co/Qwen/Qwen2.5-Coder-7B-Instruct) to 4-bit using bitsandbytes. For more information about the model,
|
8 |
+
refer to the model's page.
|
9 |
+
|
10 |
+
## Impact on performance
|
11 |
+
Impact of quantization on a set of models.
|
12 |
+
|
13 |
+
Evaluation of the model was conducted using the **PoLL (Pool of LLM)** technique, focusing on **100 French programming and computer science questions** spanning
|
14 |
+
topics in pseudo-code, Java, Python, and JavaScript. Performance was assessed through scores aggregated from **six evaluations** (two per evaluator). The evaluators
|
15 |
+
included GPT-4o, Gemini-1.5-pro, and Claude3.5-sonnet, each analyzing the model’s understanding and response accuracy on code syntax, logic, and technical
|
16 |
+
explanations across these languages and formats.
|
17 |
+
|
18 |
+
Performance Scores (on a scale of 5):
|
19 |
+
| Model | Score | # params (Billion) | size (GB) |
|
20 |
+
|:----------------------------------------------------------|:--------:|:------------------:|:---------:|
|
21 |
+
| ('Qwen/Qwen2.5-Coder-1.5B', 'Java') | 0.545072 |\n| ('Qwen/Qwen2.5-Coder-1.5B', 'JavaScript') | 0.436294 |\n| ('Qwen/Qwen2.5-Coder-1.5B', 'Python') | 0.434845 |\n| ('bigcode/starcoder2-15b', 'Java') | 0.62873 |\n| ('bigcode/starcoder2-15b', 'JavaScript') | 0.536091 |\n| ('bigcode/starcoder2-15b', 'Python') | 0.510945 |\n| ('bigcode/starcoder2-3b', 'Java') | 0.595437 |\n| ('bigcode/starcoder2-3b', 'JavaScript') | 0.500828 |\n| ('bigcode/starcoder2-3b', 'Python') | 0.479801 |\n| ('bigcode/starcoder2-7b', 'Java') | 0.574745 |\n| ('bigcode/starcoder2-7b', 'JavaScript') | 0.465017 |\n| ('bigcode/starcoder2-7b', 'Python') | 0.453512 |\n| ('cmarkea/CodeLlama-34b-hf-4bit', 'Java') | 0.724881 |\n| ('cmarkea/CodeLlama-34b-hf-4bit', 'JavaScript') | 0.609453 |\n| ('cmarkea/CodeLlama-34b-hf-4bit', 'Python') | 0.596516 |\n| ('cmarkea/CodeLlama-70b-hf-4bit', 'Java') | 0.760241 |\n| ('cmarkea/CodeLlama-70b-hf-4bit', 'JavaScript') | 0.652255 |\n| ('cmarkea/CodeLlama-70b-hf-4bit', 'Python') | 0.640797 |\n| ('cmarkea/CodeLlama-7b-hf-4bit', 'Java') | 0.66728 |\n| ('cmarkea/CodeLlama-7b-hf-4bit', 'JavaScript') | 0.542181 |\n| ('cmarkea/CodeLlama-7b-hf-4bit', 'Python') | 0.529498 |\n| ('deepseek-ai/deepseek-coder-1.3b-base', 'Java') | 0.586311 |\n| ('deepseek-ai/deepseek-coder-1.3b-base', 'JavaScript') | 0.450518 |\n| ('deepseek-ai/deepseek-coder-1.3b-base', 'Python') | 0.487962 |\n| ('deepseek-ai/deepseek-coder-6.7b-base', 'Java') | 0.6271 |\n| ('deepseek-ai/deepseek-coder-6.7b-base', 'JavaScript') | 0.496296 |\n| ('deepseek-ai/deepseek-coder-6.7b-base', 'Python') | 0.532503 |\n| ('deepseek-ai/deepseek-coder-7b-base-v1.5', 'Java') | 0.623582 |\n| ('deepseek-ai/deepseek-coder-7b-base-v1.5', 'JavaScript') | 0.503935 |\n| ('deepseek-ai/deepseek-coder-7b-base-v1.5', 'Python') | 0.521639 |\n| ('meta-llama/CodeLlama-13b-hf', 'Java') | 0.680939 |\n| ('meta-llama/CodeLlama-13b-hf', 'JavaScript') | 0.563145 |\n| ('meta-llama/CodeLlama-13b-hf', 'Python') | 0.546593 |\n| ('meta-llama/CodeLlama-34b-hf', 'Java') | 0.728144 |\n| ('meta-llama/CodeLlama-34b-hf', 'JavaScript') | 0.613087 |\n| ('meta-llama/CodeLlama-34b-hf', 'Python') | 0.599918 |\n| ('meta-llama/CodeLlama-70b-hf', 'Java') | 0.76577 |\n| ('meta-llama/CodeLlama-70b-hf', 'JavaScript') | 0.658665 |\n| ('meta-llama/CodeLlama-70b-hf', 'Python') | 0.647235 |\n| ('meta-llama/CodeLlama-7b-hf', 'Java') | 0.670398 |\n| ('meta-llama/CodeLlama-7b-hf', 'JavaScript') | 0.546067 |\n| ('meta-llama/CodeLlama-7b-hf', 'Python') | 0.533356 |\n| ('mistralai/Codestral-22B-v0.1', 'Java') | 0.631256 |\n| ('mistralai/Codestral-22B-v0.1', 'JavaScript') | 0.553193 |\n| ('mistralai/Codestral-22B-v0.1', 'Python') | 0.544975 |
|
22 |
+
|
23 |
+
| Model | Score | # params (Billion) | size (GB) |
|
24 |
+
|---------------------------------------------:|:--------:|:------------------:|:---------:|
|
25 |
+
| gpt-4o | 4.51 | N/A | N/A |
|
26 |
+
| deepseek-ai/DeepSeek-Coder-V2-Lite-Instruct | 4.24 | 15.7 | 31.4 |
|
27 |
+
| meta-llama/Meta-Llama-3.1-70B-Instruct | 4.23 | 70.06 | 141.2 |
|
28 |
+
| cmarkea/Meta-Llama-3.1-70B-Instruct-4bit | 4.14 | 70.06 | 35.3 |
|
29 |
+
| Qwen/Qwen2.5-Coder-7B-Instruct | 4.11 | 7.62 | 15.24 |
|
30 |
+
| **cmarkea/Qwen2.5-Coder-7B-Instruct-4bit** | **4.08** | **7.62** | **3.81** |
|
31 |
+
| cmarkea/Mixtral-8x7B-Instruct-v0.1-4bit | 3.8 | 46.7 | 23.35 |
|
32 |
+
| meta-llama/Meta-Llama-3.1-8B-Instruct | 3.73 | 8.03 | 16.06 |
|
33 |
+
| mistralai/Mixtral-8x7B-Instruct-v0.1 | 3.33 | 46.7 | 93.4 |
|
34 |
+
| codellama/CodeLlama-13b-Instruct-hf | 3.33 | 13 | 26 |
|
35 |
+
| codellama/CodeLlama-34b-Instruct-hf | 3.27 | 33.7 | 67.4 |
|
36 |
+
| codellama/CodeLlama-7b-Instruct-hf | 3.19 | 6.74 | 13.48 |
|
37 |
+
| cmarkea/CodeLlama-34b-Instruct-hf-4bit | 3.12 | 33.7 | 16.35 |
|
38 |
+
| codellama/CodeLlama-70b-Instruct-hf | 1.82 | 69 | 138 |
|
39 |
+
| cmarkea/CodeLlama-70b-Instruct-hf-4bit | 1.64 | 69 | 34.5 |
|
40 |
+
|
41 |
+
The impact of quantization is negligible.
|
42 |
+
|
43 |
+
## Prompt Pattern
|
44 |
+
Here is a reminder of the command pattern to interact with the model:
|
45 |
+
```verbatim
|
46 |
+
<|im_start|>user\n{user_prompt_1}<|im_end|><|im_start|>assistant\n{model_answer_1}<|im_end|>...
|
47 |
+
```
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|