ken123777 commited on
Commit
93cc4fb
Β·
verified Β·
1 Parent(s): 9923ab6

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +43 -34
README.md CHANGED
@@ -1,19 +1,19 @@
1
  ---
2
  license: mit
3
  language:
4
- - en
5
- - ko
6
- - code
7
  library_name: transformers
8
  tags:
9
- - code-llama
10
- - code-review
11
- - fine-tuning
12
- - SFT
13
- - LoRA
14
  pipeline_tag: text-generation
15
  base_model:
16
- - codellama/CodeLlama-7b-hf
17
  ---
18
 
19
  # Model Card for codellama-7b-code-review
@@ -23,11 +23,11 @@ base_model:
23
  ## Model Details / λͺ¨λΈ 상세 정보
24
 
25
  <details>
26
- <summary><strong>English</strong></summary>
27
 
28
  This model is fine-tuned from Meta's `codellama/CodeLlama-7b-hf` to review and provide feedback on code changes (`diffs`) from GitHub Pull Requests. It has been primarily trained on JavaScript and React code reviews, aiming to generate constructive feedback from a senior engineer's perspective on topics like code quality, architecture, performance, and conventions.
29
 
30
- - **Developed by:** [ken123777](https://huggingface.co/ken123777)
31
  - **Model type:** Causal Language Model
32
  - **Language(s):** English, Korean, Diff format
33
  - **License:** apache-2.0
@@ -36,11 +36,11 @@ This model is fine-tuned from Meta's `codellama/CodeLlama-7b-hf` to review and p
36
  </details>
37
 
38
  <details>
39
- <summary><strong>ν•œκ΅­μ–΄</strong></summary>
40
 
41
  이 λͺ¨λΈμ€ Meta의 `codellama/CodeLlama-7b-hf` λͺ¨λΈμ„ 기반으둜, GitHub Pull Request의 μ½”λ“œ 변경사항(`diff`)을 λ¦¬λ·°ν•˜κ³  ν”Όλ“œλ°±μ„ μ œκ³΅ν•˜λ„λ‘ νŒŒμΈνŠœλ‹λ˜μ—ˆμŠ΅λ‹ˆλ‹€. 주둜 JavaScript와 React μ½”λ“œ 리뷰에 쀑점을 두고 ν•™μŠ΅λ˜μ—ˆμœΌλ©°, μ‹œλ‹ˆμ–΄ μ—”μ§€λ‹ˆμ–΄μ˜ κ΄€μ μ—μ„œ μ½”λ“œ ν’ˆμ§ˆ, μ•„ν‚€ν…μ²˜, μ„±λŠ₯, μ»¨λ²€μ…˜ 등에 λŒ€ν•œ 건섀적인 ν”Όλ“œλ°±μ„ μƒμ„±ν•˜λŠ” 것을 λͺ©ν‘œλ‘œ ν•©λ‹ˆλ‹€.
42
 
43
- - **개발자:** [ken123777](https://huggingface.co/ken123777)
44
  - **λͺ¨λΈ μ’…λ₯˜:** 인과 관계 μ–Έμ–΄ λͺ¨λΈ (Causal Language Model)
45
  - **μ–Έμ–΄:** μ˜μ–΄, ν•œκ΅­μ–΄, Diff ν˜•μ‹
46
  - **λΌμ΄μ„ μŠ€:** apache-2.0
@@ -55,7 +55,7 @@ This model is fine-tuned from Meta's `codellama/CodeLlama-7b-hf` to review and p
55
  ## Uses / μ‚¬μš© 정보
56
 
57
  <details>
58
- <summary><strong>English</strong></summary>
59
 
60
  ### Direct Use
61
 
@@ -74,7 +74,7 @@ This model is specialized for code review tasks. It may not perform well for oth
74
  </details>
75
 
76
  <details>
77
- <summary><strong>ν•œκ΅­μ–΄</strong></summary>
78
 
79
  ### 직접 μ‚¬μš©
80
 
@@ -95,7 +95,7 @@ This model is specialized for code review tasks. It may not perform well for oth
95
  ## Bias, Risks, and Limitations / 편ν–₯, μœ„ν—˜ 및 ν•œκ³„
96
 
97
  <details>
98
- <summary><strong>English</strong></summary>
99
 
100
  - **Data Bias:** The model was trained on public GitHub Pull Request data, so it may be biased towards specific coding styles or patterns present in that data.
101
  - **Inaccuracy (Hallucination):** The model may occasionally generate feedback that is factually incorrect or out of context. The generated reviews always need verification.
@@ -103,7 +103,7 @@ This model is specialized for code review tasks. It may not perform well for oth
103
  </details>
104
 
105
  <details>
106
- <summary><strong>ν•œκ΅­μ–΄</strong></summary>
107
 
108
  - **데이터 편ν–₯:** λͺ¨λΈμ€ 곡개된 GitHub Pull Request 데이터λ₯Ό 기반으둜 ν•™μŠ΅λ˜μ—ˆμœΌλ―€λ‘œ, ν•΄λ‹Ή 데이터에 μ‘΄μž¬ν•˜λŠ” νŠΉμ • μ½”λ”© μŠ€νƒ€μΌμ΄λ‚˜ νŒ¨ν„΄μ— 편ν–₯λ˜μ–΄ μžˆμ„ 수 μžˆμŠ΅λ‹ˆλ‹€.
109
  - **λΆ€μ •ν™•μ„±(ν™˜κ°):** λͺ¨λΈμ€ λ•Œλ•Œλ‘œ 사싀과 λ‹€λ₯΄κ±°λ‚˜ λ¬Έλ§₯에 λ§žμ§€ μ•ŠλŠ” ν”Όλ“œλ°±μ„ 생성할 수 μžˆμŠ΅λ‹ˆλ‹€. μƒμ„±λœ λ¦¬λ·°λŠ” 항상 검증이 ν•„μš”ν•©λ‹ˆλ‹€.
@@ -113,19 +113,19 @@ This model is specialized for code review tasks. It may not perform well for oth
113
  ### Recommendations / ꢌμž₯ 사항
114
 
115
  <details>
116
- <summary><strong>English</strong></summary>
117
  Users should treat the code reviews generated by the model as a 'draft' or 'assistive tool' to help the development process, not as a final judgment. It is recommended that a human expert reviews critical changes.
118
  </details>
119
 
120
  <details>
121
- <summary><strong>ν•œκ΅­μ–΄</strong></summary>
122
  μ‚¬μš©μžλŠ” λͺ¨λΈμ΄ μƒμ„±ν•œ μ½”λ“œ 리뷰λ₯Ό μ΅œμ’…μ μΈ νŒλ‹¨μ΄ μ•„λ‹Œ, 개발 과정을 λ•λŠ” 'μ΄ˆμ•ˆ' λ˜λŠ” '보쑰 도ꡬ'둜 ν™œμš©ν•΄μ•Ό ν•©λ‹ˆλ‹€. μ€‘μš”ν•œ 변경사항에 λŒ€ν•΄μ„œλŠ” λ°˜λ“œμ‹œ 인간 μ „λ¬Έκ°€μ˜ κ²€ν† λ₯Ό κ±°μΉ˜λŠ” 것을 ꢌμž₯ν•©λ‹ˆλ‹€.
123
  </details>
124
 
125
  ## How to Get Started with the Model / λͺ¨λΈ μ‹œμž‘ν•˜κΈ°
126
 
127
  <details>
128
- <summary><strong>English</strong></summary>
129
 
130
  **Note:** This model may be available in two versions: **Adapter** and **Merged**. Use the appropriate code for your model type.
131
 
@@ -140,15 +140,15 @@ If the model is fully merged with the base model, you can load it directly witho
140
  </details>
141
 
142
  <details>
143
- <summary><strong>ν•œκ΅­μ–΄</strong></summary>
144
 
145
  **μ°Έκ³ :** 이 λͺ¨λΈμ€ **μ–΄λŒ‘ν„°(Adapter)** 와 **λ³‘ν•©λœ(Merged)** 두 κ°€μ§€ λ²„μ „μœΌλ‘œ 제곡될 수 μžˆμŠ΅λ‹ˆλ‹€. μžμ‹ μ˜ λͺ¨λΈ νƒ€μž…μ— λ§žλŠ” μ½”λ“œλ₯Ό μ‚¬μš©ν•˜μ„Έμš”.
146
 
147
- #### 1. μ–΄λŒ‘ν„° λͺ¨λΈ μ‚¬μš©λ²• (`ken123777/codellama-7b-code-review-adapter`)
148
 
149
  μ–΄λŒ‘ν„° λͺ¨λΈμ„ μ‚¬μš©ν•˜λ €λ©΄, 기반 λͺ¨λΈμ„ λ¨Όμ € λ‘œλ“œν•œ ν›„ `peft` 라이브러리λ₯Ό μ‚¬μš©ν•΄ μ–΄λŒ‘ν„°λ₯Ό μ μš©ν•΄μ•Ό ν•©λ‹ˆλ‹€.
150
 
151
- #### 2. λ³‘ν•©λœ λͺ¨λΈ μ‚¬μš©λ²• (`ken123777/codellama-7b-code-review`)
152
 
153
  λͺ¨λΈμ΄ 기반 λͺ¨λΈκ³Ό μ™„μ „νžˆ λ³‘ν•©λœ 경우, `peft` 없이 직접 λͺ¨λΈμ„ λ‘œλ“œν•˜μ—¬ μ‚¬μš©ν•  수 μžˆμŠ΅λ‹ˆλ‹€.
154
 
@@ -213,39 +213,42 @@ diff_code = """
213
  """
214
 
215
  # Prompt in Korean
 
 
216
  prompt = f"""### μ§€μ‹œ:
217
  제곡된 μ½”λ“œλŠ” pull request의 diff λ‚΄μš©μž…λ‹ˆλ‹€. μ½”λ“œμ˜ κ°œμ„ ν•  수 μžˆλŠ” 뢀뢄에 λŒ€ν•΄ μ΅œμ†Œ 3κ°€μ§€ ν•­λͺ©μœΌλ‘œ λ‚˜λˆ„μ–΄ μƒμ„Έν•˜κ³  ꡬ체적인 ν”Όλ“œλ°±μ„ μ œκ³΅ν•΄μ£Όμ„Έμš”.
218
 
219
  ### μž…λ ₯:
220
- ```diff
221
  {diff_code}
222
- ````
223
 
224
  ### 응닡:
225
-
226
  1. """
227
 
228
  inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
229
- outputs = model.generate(\*\*inputs, max_new_tokens=512, temperature=0.7, repetition_penalty=1.2)
230
  response = tokenizer.decode(outputs[0]len(inputs.input_ids[0]):], skip_special_tokens=True)
231
 
232
  print(response)
233
 
234
- ```
235
-
236
 
237
  ## Training Details / ν•™μŠ΅ 상세 정보
238
 
239
  <details>
240
- <summary><strong>English</strong></summary>
241
 
242
  ### Training Data
 
243
  This model was fine-tuned using the `review_dataset.json` file, which contains public Pull Request data collected from GitHub. The dataset is structured in a `instruction`, `input`(diff), `output`(review comment) format.
244
 
245
  ### Training Procedure
 
246
  The model was fine-tuned using the QLoRA technique. It utilized the `SFTTrainer` from the `trl` library, applying 4-bit quantization and LoRA (Low-Rank Adaptation) for efficient training.
247
 
248
  #### Training Hyperparameters
 
249
  - **model:** `codellama/CodeLlama-7b-hf`
250
  - **max_seq_length:** 4096
251
  - **lora_alpha:** 128
@@ -260,15 +263,18 @@ The model was fine-tuned using the QLoRA technique. It utilized the `SFTTrainer`
260
  </details>
261
 
262
  <details>
263
- <summary><strong>ν•œκ΅­μ–΄</strong></summary>
264
 
265
  ### ν•™μŠ΅ 데이터
 
266
  이 λͺ¨λΈμ€ GitHubμ—μ„œ μˆ˜μ§‘λœ 곡개 Pull Request 데이터λ₯Ό ν¬ν•¨ν•˜λŠ” `review_dataset.json` νŒŒμΌμ„ μ‚¬μš©ν•˜μ—¬ νŒŒμΈνŠœλ‹λ˜μ—ˆμŠ΅λ‹ˆλ‹€. 데이터셋은 `instruction`, `input`(diff), `output`(리뷰 μ½”λ©˜νŠΈ) ν˜•μ‹μœΌλ‘œ κ΅¬μ„±λ˜μ–΄ μžˆμŠ΅λ‹ˆλ‹€.
267
 
268
  ### ν•™μŠ΅ 절차
 
269
  λͺ¨λΈμ€ QLoRA 기법을 μ‚¬μš©ν•˜μ—¬ νŒŒμΈνŠœλ‹λ˜μ—ˆμŠ΅λ‹ˆλ‹€. `trl` 라이브러리의 `SFTTrainer`λ₯Ό μ‚¬μš©ν–ˆμœΌλ©°, 4-bit μ–‘μžν™”μ™€ LoRA(Low-Rank Adaptation)λ₯Ό μ μš©ν•˜μ—¬ 효율적인 ν•™μŠ΅μ„ μ§„ν–‰ν–ˆμŠ΅λ‹ˆλ‹€.
270
 
271
  #### ν•™μŠ΅ ν•˜μ΄νΌνŒŒλΌλ―Έν„°
 
272
  - **λͺ¨λΈ:** `codellama/CodeLlama-7b-hf`
273
  - **μ΅œλŒ€ μ‹œν€€μŠ€ 길이:** 4096
274
  - **LoRA Alpha:** 128
@@ -285,16 +291,19 @@ The model was fine-tuned using the QLoRA technique. It utilized the `SFTTrainer`
285
  ## Compute Infrastructure / μ»΄ν“¨νŒ… 인프라
286
 
287
  <details>
288
- <summary><strong>English</strong></summary>
289
 
290
  - **Hardware Type:** RunPod Cloud GPU
291
  - **Cloud Provider:** RunPod
292
  </details>
293
 
294
  <details>
295
- <summary><strong>ν•œκ΅­μ–΄</strong></summary>
296
 
297
  - **ν•˜λ“œμ›¨μ–΄ μ’…λ₯˜:** RunPod ν΄λΌμš°λ“œ GPU
298
  - **ν΄λΌμš°λ“œ μ œκ³΅μ—…μ²΄:** RunPod
299
  </details>
300
- ```
 
 
 
 
1
  ---
2
  license: mit
3
  language:
4
+ - en
5
+ - ko
6
+ - code
7
  library_name: transformers
8
  tags:
9
+ - code-llama
10
+ - code-review
11
+ - fine-tuning
12
+ - SFT
13
+ - LoRA
14
  pipeline_tag: text-generation
15
  base_model:
16
+ - codellama/CodeLlama-7b-hf
17
  ---
18
 
19
  # Model Card for codellama-7b-code-review
 
23
  ## Model Details / λͺ¨λΈ 상세 정보
24
 
25
  <details>
26
+ <summary><strong>πŸ‡ΊπŸ‡Έ English</strong></summary>
27
 
28
  This model is fine-tuned from Meta's `codellama/CodeLlama-7b-hf` to review and provide feedback on code changes (`diffs`) from GitHub Pull Requests. It has been primarily trained on JavaScript and React code reviews, aiming to generate constructive feedback from a senior engineer's perspective on topics like code quality, architecture, performance, and conventions.
29
 
30
+ - **Developed by:** [ken12377](https://huggingface.co/ken12377)
31
  - **Model type:** Causal Language Model
32
  - **Language(s):** English, Korean, Diff format
33
  - **License:** apache-2.0
 
36
  </details>
37
 
38
  <details>
39
+ <summary><strong>πŸ‡°πŸ‡· ν•œκ΅­μ–΄</strong></summary>
40
 
41
  이 λͺ¨λΈμ€ Meta의 `codellama/CodeLlama-7b-hf` λͺ¨λΈμ„ 기반으둜, GitHub Pull Request의 μ½”λ“œ 변경사항(`diff`)을 λ¦¬λ·°ν•˜κ³  ν”Όλ“œλ°±μ„ μ œκ³΅ν•˜λ„λ‘ νŒŒμΈνŠœλ‹λ˜μ—ˆμŠ΅λ‹ˆλ‹€. 주둜 JavaScript와 React μ½”λ“œ 리뷰에 쀑점을 두고 ν•™μŠ΅λ˜μ—ˆμœΌλ©°, μ‹œλ‹ˆμ–΄ μ—”μ§€λ‹ˆμ–΄μ˜ κ΄€μ μ—μ„œ μ½”λ“œ ν’ˆμ§ˆ, μ•„ν‚€ν…μ²˜, μ„±λŠ₯, μ»¨λ²€μ…˜ 등에 λŒ€ν•œ 건섀적인 ν”Όλ“œλ°±μ„ μƒμ„±ν•˜λŠ” 것을 λͺ©ν‘œλ‘œ ν•©λ‹ˆλ‹€.
42
 
43
+ - **개발자:** [ken12377](https://huggingface.co/ken12377)
44
  - **λͺ¨λΈ μ’…λ₯˜:** 인과 관계 μ–Έμ–΄ λͺ¨λΈ (Causal Language Model)
45
  - **μ–Έμ–΄:** μ˜μ–΄, ν•œκ΅­μ–΄, Diff ν˜•μ‹
46
  - **λΌμ΄μ„ μŠ€:** apache-2.0
 
55
  ## Uses / μ‚¬μš© 정보
56
 
57
  <details>
58
+ <summary><strong>πŸ‡ΊπŸ‡Έ English</strong></summary>
59
 
60
  ### Direct Use
61
 
 
74
  </details>
75
 
76
  <details>
77
+ <summary><strong>πŸ‡°πŸ‡· ν•œκ΅­μ–΄</strong></summary>
78
 
79
  ### 직접 μ‚¬μš©
80
 
 
95
  ## Bias, Risks, and Limitations / 편ν–₯, μœ„ν—˜ 및 ν•œκ³„
96
 
97
  <details>
98
+ <summary><strong>πŸ‡ΊπŸ‡Έ English</strong></summary>
99
 
100
  - **Data Bias:** The model was trained on public GitHub Pull Request data, so it may be biased towards specific coding styles or patterns present in that data.
101
  - **Inaccuracy (Hallucination):** The model may occasionally generate feedback that is factually incorrect or out of context. The generated reviews always need verification.
 
103
  </details>
104
 
105
  <details>
106
+ <summary><strong>πŸ‡°πŸ‡· ν•œκ΅­μ–΄</strong></summary>
107
 
108
  - **데이터 편ν–₯:** λͺ¨λΈμ€ 곡개된 GitHub Pull Request 데이터λ₯Ό 기반으둜 ν•™μŠ΅λ˜μ—ˆμœΌλ―€λ‘œ, ν•΄λ‹Ή 데이터에 μ‘΄μž¬ν•˜λŠ” νŠΉμ • μ½”λ”© μŠ€νƒ€μΌμ΄λ‚˜ νŒ¨ν„΄μ— 편ν–₯λ˜μ–΄ μžˆμ„ 수 μžˆμŠ΅λ‹ˆλ‹€.
109
  - **λΆ€μ •ν™•μ„±(ν™˜κ°):** λͺ¨λΈμ€ λ•Œλ•Œλ‘œ 사싀과 λ‹€λ₯΄κ±°λ‚˜ λ¬Έλ§₯에 λ§žμ§€ μ•ŠλŠ” ν”Όλ“œλ°±μ„ 생성할 수 μžˆμŠ΅λ‹ˆλ‹€. μƒμ„±λœ λ¦¬λ·°λŠ” 항상 검증이 ν•„μš”ν•©λ‹ˆλ‹€.
 
113
  ### Recommendations / ꢌμž₯ 사항
114
 
115
  <details>
116
+ <summary><strong>πŸ‡ΊπŸ‡Έ English</strong></summary>
117
  Users should treat the code reviews generated by the model as a 'draft' or 'assistive tool' to help the development process, not as a final judgment. It is recommended that a human expert reviews critical changes.
118
  </details>
119
 
120
  <details>
121
+ <summary><strong>πŸ‡°πŸ‡· ν•œκ΅­μ–΄</strong></summary>
122
  μ‚¬μš©μžλŠ” λͺ¨λΈμ΄ μƒμ„±ν•œ μ½”λ“œ 리뷰λ₯Ό μ΅œμ’…μ μΈ νŒλ‹¨μ΄ μ•„λ‹Œ, 개발 과정을 λ•λŠ” 'μ΄ˆμ•ˆ' λ˜λŠ” '보쑰 도ꡬ'둜 ν™œμš©ν•΄μ•Ό ν•©λ‹ˆλ‹€. μ€‘μš”ν•œ 변경사항에 λŒ€ν•΄μ„œλŠ” λ°˜λ“œμ‹œ 인간 μ „λ¬Έκ°€μ˜ κ²€ν† λ₯Ό κ±°μΉ˜λŠ” 것을 ꢌμž₯ν•©λ‹ˆλ‹€.
123
  </details>
124
 
125
  ## How to Get Started with the Model / λͺ¨λΈ μ‹œμž‘ν•˜κΈ°
126
 
127
  <details>
128
+ <summary><strong>πŸ‡ΊπŸ‡Έ English</strong></summary>
129
 
130
  **Note:** This model may be available in two versions: **Adapter** and **Merged**. Use the appropriate code for your model type.
131
 
 
140
  </details>
141
 
142
  <details>
143
+ <summary><strong>πŸ‡°πŸ‡· ν•œκ΅­μ–΄</strong></summary>
144
 
145
  **μ°Έκ³ :** 이 λͺ¨λΈμ€ **μ–΄λŒ‘ν„°(Adapter)** 와 **λ³‘ν•©λœ(Merged)** 두 κ°€μ§€ λ²„μ „μœΌλ‘œ 제곡될 수 μžˆμŠ΅λ‹ˆλ‹€. μžμ‹ μ˜ λͺ¨λΈ νƒ€μž…μ— λ§žλŠ” μ½”λ“œλ₯Ό μ‚¬μš©ν•˜μ„Έμš”.
146
 
147
+ #### 1. μ–΄λŒ‘ν„° λͺ¨λΈ μ‚¬μš©λ²• (`ken12377/codellama-7b-code-review-adapter`)
148
 
149
  μ–΄λŒ‘ν„° λͺ¨λΈμ„ μ‚¬μš©ν•˜λ €λ©΄, 기반 λͺ¨λΈμ„ λ¨Όμ € λ‘œλ“œν•œ ν›„ `peft` 라이브러리λ₯Ό μ‚¬μš©ν•΄ μ–΄λŒ‘ν„°λ₯Ό μ μš©ν•΄μ•Ό ν•©λ‹ˆλ‹€.
150
 
151
+ #### 2. λ³‘ν•©λœ λͺ¨λΈ μ‚¬μš©λ²• (`ken12377/codellama-7b-code-review`)
152
 
153
  λͺ¨λΈμ΄ 기반 λͺ¨λΈκ³Ό μ™„μ „νžˆ λ³‘ν•©λœ 경우, `peft` 없이 직접 λͺ¨λΈμ„ λ‘œλ“œν•˜μ—¬ μ‚¬μš©ν•  수 μžˆμŠ΅λ‹ˆλ‹€.
154
 
 
213
  """
214
 
215
  # Prompt in Korean
216
+ # λ§ˆν¬λ‹€μš΄ νŒŒμ„œμ˜ ν˜Όλ™μ„ ν”Όν•˜κΈ° μœ„ν•΄ μ½”λ“œ 블둝 κ΅¬λΆ„μžλ₯Ό λ³€μˆ˜λ‘œ λ§Œλ“€μ–΄ μ‚¬μš©ν•©λ‹ˆλ‹€.
217
+ diff_block_delimiter = "```"
218
  prompt = f"""### μ§€μ‹œ:
219
  제곡된 μ½”λ“œλŠ” pull request의 diff λ‚΄μš©μž…λ‹ˆλ‹€. μ½”λ“œμ˜ κ°œμ„ ν•  수 μžˆλŠ” 뢀뢄에 λŒ€ν•΄ μ΅œμ†Œ 3κ°€μ§€ ν•­λͺ©μœΌλ‘œ λ‚˜λˆ„μ–΄ μƒμ„Έν•˜κ³  ꡬ체적인 ν”Όλ“œλ°±μ„ μ œκ³΅ν•΄μ£Όμ„Έμš”.
220
 
221
  ### μž…λ ₯:
222
+ {diff_block_delimiter}diff
223
  {diff_code}
224
+ {diff_block_delimiter}
225
 
226
  ### 응닡:
 
227
  1. """
228
 
229
  inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
230
+ outputs = model.generate(**inputs, max_new_tokens=512, temperature=0.7, repetition_penalty=1.2)
231
  response = tokenizer.decode(outputs[0]len(inputs.input_ids[0]):], skip_special_tokens=True)
232
 
233
  print(response)
234
 
235
+ ````
 
236
 
237
  ## Training Details / ν•™μŠ΅ 상세 정보
238
 
239
  <details>
240
+ <summary><strong>πŸ‡ΊπŸ‡Έ English</strong></summary>
241
 
242
  ### Training Data
243
+
244
  This model was fine-tuned using the `review_dataset.json` file, which contains public Pull Request data collected from GitHub. The dataset is structured in a `instruction`, `input`(diff), `output`(review comment) format.
245
 
246
  ### Training Procedure
247
+
248
  The model was fine-tuned using the QLoRA technique. It utilized the `SFTTrainer` from the `trl` library, applying 4-bit quantization and LoRA (Low-Rank Adaptation) for efficient training.
249
 
250
  #### Training Hyperparameters
251
+
252
  - **model:** `codellama/CodeLlama-7b-hf`
253
  - **max_seq_length:** 4096
254
  - **lora_alpha:** 128
 
263
  </details>
264
 
265
  <details>
266
+ <summary><strong>πŸ‡°πŸ‡· ν•œκ΅­μ–΄</strong></summary>
267
 
268
  ### ν•™μŠ΅ 데이터
269
+
270
  이 λͺ¨λΈμ€ GitHubμ—μ„œ μˆ˜μ§‘λœ 곡개 Pull Request 데이터λ₯Ό ν¬ν•¨ν•˜λŠ” `review_dataset.json` νŒŒμΌμ„ μ‚¬μš©ν•˜μ—¬ νŒŒμΈνŠœλ‹λ˜μ—ˆμŠ΅λ‹ˆλ‹€. 데이터셋은 `instruction`, `input`(diff), `output`(리뷰 μ½”λ©˜νŠΈ) ν˜•μ‹μœΌλ‘œ κ΅¬μ„±λ˜μ–΄ μžˆμŠ΅λ‹ˆλ‹€.
271
 
272
  ### ν•™μŠ΅ 절차
273
+
274
  λͺ¨λΈμ€ QLoRA 기법을 μ‚¬μš©ν•˜μ—¬ νŒŒμΈνŠœλ‹λ˜μ—ˆμŠ΅λ‹ˆλ‹€. `trl` 라이브러리의 `SFTTrainer`λ₯Ό μ‚¬μš©ν–ˆμœΌλ©°, 4-bit μ–‘μžν™”μ™€ LoRA(Low-Rank Adaptation)λ₯Ό μ μš©ν•˜μ—¬ 효율적인 ν•™μŠ΅μ„ μ§„ν–‰ν–ˆμŠ΅λ‹ˆλ‹€.
275
 
276
  #### ν•™μŠ΅ ν•˜μ΄νΌνŒŒλΌλ―Έν„°
277
+
278
  - **λͺ¨λΈ:** `codellama/CodeLlama-7b-hf`
279
  - **μ΅œλŒ€ μ‹œν€€μŠ€ 길이:** 4096
280
  - **LoRA Alpha:** 128
 
291
  ## Compute Infrastructure / μ»΄ν“¨νŒ… 인프라
292
 
293
  <details>
294
+ <summary><strong>πŸ‡ΊπŸ‡Έ English</strong></summary>
295
 
296
  - **Hardware Type:** RunPod Cloud GPU
297
  - **Cloud Provider:** RunPod
298
  </details>
299
 
300
  <details>
301
+ <summary><strong>πŸ‡°πŸ‡· ν•œκ΅­μ–΄</strong></summary>
302
 
303
  - **ν•˜λ“œμ›¨μ–΄ μ’…λ₯˜:** RunPod ν΄λΌμš°λ“œ GPU
304
  - **ν΄λΌμš°λ“œ μ œκ³΅μ—…μ²΄:** RunPod
305
  </details>
306
+
307
+ ```
308
+
309
+ ```