nayohan commited on
Commit
04620d1
β€’
1 Parent(s): 101d48e

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +5 -5
README.md CHANGED
@@ -105,7 +105,7 @@ OUTPUT: 기술 및 기초 과학은 연ꡬ μΈ‘λ©΄μ—μ„œ 맀우 μ€‘μš”ν•©λ‹ˆλ‹€.
105
  <br>
106
 
107
  ## **Aihub 영-ν•œ λ²ˆμ—­λ°μ΄ν„°μ…‹ 평가**
108
- * Aihub 평가 데이터셋은 λͺ¨λΈλ“€μ΄ ν•™μŠ΅λ°μ΄ν„°μ…‹μ— ν¬ν•¨λ˜μ—ˆμ„ 수 μžˆμŠ΅λ‹ˆλ‹€. μΉ΄ν…Œκ³ λ¦¬λ³„ μ„±λŠ₯을 ν™•μΈν•˜λŠ” μš©λ„λ‘œλ§Œ μ°Έκ³ ν•΄μ£Όμ„Έμš”.
109
  ![image/png](https://cdn-uploads.huggingface.co/production/uploads/6152b4b9ecf3ca6ab820e325/TMo05LOUhPGYNbT2ADOgi.png)
110
  | model | aihub-111 | aihub-124 | aihub-125 | aihub-126 | aihub-563 | aihub-71265 | aihub-71266 | aihub-71382 | average |
111
  |:-----------------|------------:|------------:|------------:|------------:|------------:|--------------:|--------------:|--------------:|----------:|
@@ -120,7 +120,7 @@ OUTPUT: 기술 및 기초 과학은 연ꡬ μΈ‘λ©΄μ—μ„œ 맀우 μ€‘μš”ν•©λ‹ˆλ‹€.
120
  | our-instrucTrans | 24.89 | 47.00 | 22.78 | 21.78 | 24.27 | 27.98 | 31.31 | 15.42 |**26.92** |
121
  ## **FLoRes 영-ν•œ λ²ˆμ—­λ°μ΄ν„°μ…‹ 평가**
122
  [FloRes](https://huggingface.co/datasets/facebook/flores)λŠ” νŽ˜μ΄μŠ€λΆμ—μ„œ κ³΅κ°œν•œ μ˜μ–΄μ™€ 적은 λ¦¬μ†ŒμŠ€μ˜ μ–Έμ–΄ 200κ°œμ— λŒ€ν•΄μ„œ λ³‘λ ¬λ‘œ κ΅¬μ„±ν•œ λ²ˆμ—­ 벀치마크 λ°μ΄ν„°μ…‹μž…λ‹ˆλ‹€.
123
- traintogpb/aihub-flores-koen-integrated-sparta-30kλ₯Ό ν™œμš©ν•˜μ—¬ 평가λ₯Ό μ§„ν–‰ν•˜μ˜€μŠ΅λ‹ˆλ‹€. (ν•œλ¬Έμž₯ ꡬ성)
124
  ![image/png](https://cdn-uploads.huggingface.co/production/uploads/6152b4b9ecf3ca6ab820e325/ZDeA-7e-0xfXaGOmyS9zs.png)
125
  | model | flores-dev | flores-devtest | average |
126
  |:-----------------|-------------:|-----------------:|----------:|
@@ -134,9 +134,9 @@ traintogpb/aihub-flores-koen-integrated-sparta-30kλ₯Ό ν™œμš©ν•˜μ—¬ 평가λ₯Ό 진
134
  | our-sharegpt | 14.71 | 16.69 | 15.70 |
135
  | our-instrucTrans | 14.49 | 17.69 | **16.09** |
136
  ## **iwslt-2023**
137
- λ™μΌν•œ μ˜μ–΄λ¬Έμž₯을 각각 반말, μ‘΄λŒ“λ§μ˜ ν•œκ΅­μ–΄λ‘œ 평가데이터셋이 κ΅¬μ„±λ˜μ–΄ μžˆμŠ΅λ‹ˆλ‹€. λͺ¨λΈμ˜ μ‘΄λŒ€/반말 κ²½ν–₯을 μƒλŒ€μ μœΌλ‘œ 확인할 수 μžˆμŠ΅λ‹ˆλ‹€. (ν•œλ¬Έμž₯ ꡬ성)
138
  ![image/png](https://cdn-uploads.huggingface.co/production/uploads/6152b4b9ecf3ca6ab820e325/UJvuCnbjWokBWQNhD4L63.png)
139
- | model | iwlst_zondae | iwlst_banmal | average |
140
  |:-----------------|---------------------:|------------------:|----------:|
141
  | EEVE-10.8b-it | 4.62 | 3.79 | 4.20 |
142
  | KULLM3 | 5.94 | 5.24 | 5.59 |
@@ -148,7 +148,7 @@ traintogpb/aihub-flores-koen-integrated-sparta-30kλ₯Ό ν™œμš©ν•˜μ—¬ 평가λ₯Ό 진
148
  | our-sharegpt | 7.83 | 6.35 | 7.09 |
149
  | our-instrucTrans | 8.63 | 6.97 | 7.80 |
150
  ## **ko_news_eval40**
151
- 24λ…„5μ›” λ‰΄μŠ€λ₯Ό 각 μΉ΄ν…Œκ³ λ¦¬(4) 별 10κ°œμ”© 기사 λ‚΄ 문단 일뢀λ₯Ό μˆ˜μ§‘ν•˜κ³ , GPT4둜 λ²ˆμ—­ν•˜μ—¬ κ΅¬μ„±ν•˜μ˜€μŠ΅λ‹ˆλ‹€.
152
  μ˜μ–΄λ₯Ό μΌμƒλ‰΄μŠ€μ— μ‚¬μš©λ˜λŠ” ν•œκ΅­μ–΄λ‘œ 잘 λ²ˆμ—­ν•˜λŠ”μ§€λ₯Ό ν‰κ°€ν•©λ‹ˆλ‹€. (문단 ꡬ성)
153
  ![image/png](https://cdn-uploads.huggingface.co/production/uploads/6152b4b9ecf3ca6ab820e325/OaE5z_yQT9sIIz0zsn644.png)
154
  | model | IT/κ³Όν•™ | 경제 | μ‚¬νšŒ | μ˜€ν”Όλ‹ˆμ–Έ | average |
 
105
  <br>
106
 
107
  ## **Aihub 영-ν•œ λ²ˆμ—­λ°μ΄ν„°μ…‹ 평가**
108
+ * [Aihub 평가 데이터셋]](https://huggingface.co/datasets/traintogpb/aihub-flores-koen-integrated-sparta-30k)은 λͺ¨λΈλ“€μ΄ ν•™μŠ΅λ°μ΄ν„°μ…‹μ— ν¬ν•¨λ˜μ—ˆμ„ 수 μžˆμŠ΅λ‹ˆλ‹€. μΉ΄ν…Œκ³ λ¦¬λ³„ μ„±λŠ₯을 ν™•μΈν•˜λŠ” μš©λ„λ‘œλ§Œ μ°Έκ³ ν•΄μ£Όμ„Έμš”. [[μΉ΄ν…Œκ³ λ¦¬ μ„€λͺ… 링크]](https://huggingface.co/datasets/traintogpb/aihub-koen-translation-integrated-tiny-100k)
109
  ![image/png](https://cdn-uploads.huggingface.co/production/uploads/6152b4b9ecf3ca6ab820e325/TMo05LOUhPGYNbT2ADOgi.png)
110
  | model | aihub-111 | aihub-124 | aihub-125 | aihub-126 | aihub-563 | aihub-71265 | aihub-71266 | aihub-71382 | average |
111
  |:-----------------|------------:|------------:|------------:|------------:|------------:|--------------:|--------------:|--------------:|----------:|
 
120
  | our-instrucTrans | 24.89 | 47.00 | 22.78 | 21.78 | 24.27 | 27.98 | 31.31 | 15.42 |**26.92** |
121
  ## **FLoRes 영-ν•œ λ²ˆμ—­λ°μ΄ν„°μ…‹ 평가**
122
  [FloRes](https://huggingface.co/datasets/facebook/flores)λŠ” νŽ˜μ΄μŠ€λΆμ—μ„œ κ³΅κ°œν•œ μ˜μ–΄μ™€ 적은 λ¦¬μ†ŒμŠ€μ˜ μ–Έμ–΄ 200κ°œμ— λŒ€ν•΄μ„œ λ³‘λ ¬λ‘œ κ΅¬μ„±ν•œ λ²ˆμ—­ 벀치마크 λ°μ΄ν„°μ…‹μž…λ‹ˆλ‹€.
123
+ [traintogpb/aihub-flores-koen-integrated-sparta-30k](https://huggingface.co/datasets/traintogpb/aihub-flores-koen-integrated-sparta-30k)λ₯Ό ν™œμš©ν•˜μ—¬ 평가λ₯Ό μ§„ν–‰ν•˜μ˜€μŠ΅λ‹ˆλ‹€. (ν•œλ¬Έμž₯ ꡬ성)
124
  ![image/png](https://cdn-uploads.huggingface.co/production/uploads/6152b4b9ecf3ca6ab820e325/ZDeA-7e-0xfXaGOmyS9zs.png)
125
  | model | flores-dev | flores-devtest | average |
126
  |:-----------------|-------------:|-----------------:|----------:|
 
134
  | our-sharegpt | 14.71 | 16.69 | 15.70 |
135
  | our-instrucTrans | 14.49 | 17.69 | **16.09** |
136
  ## **iwslt-2023**
137
+ [iwslt-2023 데이터셋](https://huggingface.co/datasets/shreevigneshs/iwslt-2023-en-ko-train-val-split-0.1)은 λ™μΌν•œ μ˜μ–΄λ¬Έμž₯을 각각 반말, μ‘΄λŒ“λ§μ˜ ν•œκ΅­μ–΄λ‘œ 평가데이터셋이 κ΅¬μ„±λ˜μ–΄ μžˆμŠ΅λ‹ˆλ‹€. λͺ¨λΈμ˜ μ‘΄λŒ€/반말 κ²½ν–₯을 μƒλŒ€μ μœΌλ‘œ 확인할 수 μžˆμŠ΅λ‹ˆλ‹€. (ν•œλ¬Έμž₯ ꡬ성)
138
  ![image/png](https://cdn-uploads.huggingface.co/production/uploads/6152b4b9ecf3ca6ab820e325/UJvuCnbjWokBWQNhD4L63.png)
139
+ | model | iwslt_zondae | iwslt_banmal | average |
140
  |:-----------------|---------------------:|------------------:|----------:|
141
  | EEVE-10.8b-it | 4.62 | 3.79 | 4.20 |
142
  | KULLM3 | 5.94 | 5.24 | 5.59 |
 
148
  | our-sharegpt | 7.83 | 6.35 | 7.09 |
149
  | our-instrucTrans | 8.63 | 6.97 | 7.80 |
150
  ## **ko_news_eval40**
151
+ [ko_news_eval40 데이터셋](https://huggingface.co/datasets/nayohan/ko_news_eval40)은 ν•™μŠ΅λ˜μ§€ μ•Šμ•˜μ„ μƒˆλ‘œμš΄ 데이터셋에 ν‰κ°€ν•˜κ³ μž 24λ…„5μ›” λ‰΄μŠ€λ₯Ό 각 μΉ΄ν…Œκ³ λ¦¬(4) 별 10κ°œμ”© 기사 λ‚΄ 문단 일뢀λ₯Ό μˆ˜μ§‘ν•˜κ³ , GPT4둜 λ²ˆμ—­ν•˜μ—¬ κ΅¬μ„±ν•˜μ˜€μŠ΅λ‹ˆλ‹€.
152
  μ˜μ–΄λ₯Ό μΌμƒλ‰΄μŠ€μ— μ‚¬μš©λ˜λŠ” ν•œκ΅­μ–΄λ‘œ 잘 λ²ˆμ—­ν•˜λŠ”μ§€λ₯Ό ν‰κ°€ν•©λ‹ˆλ‹€. (문단 ꡬ성)
153
  ![image/png](https://cdn-uploads.huggingface.co/production/uploads/6152b4b9ecf3ca6ab820e325/OaE5z_yQT9sIIz0zsn644.png)
154
  | model | IT/κ³Όν•™ | 경제 | μ‚¬νšŒ | μ˜€ν”Όλ‹ˆμ–Έ | average |