Update README.md
Browse files
README.md
CHANGED
|
@@ -27,65 +27,139 @@ You can download the wheel built by [cognitivecomputations](https://huggingface.
|
|
| 27 |
~~~python
|
| 28 |
import requests
|
| 29 |
|
| 30 |
-
url = "http://localhost:12345/v1/completions"
|
| 31 |
headers = {"Content-Type": "application/json"}
|
|
|
|
| 32 |
data = {
|
| 33 |
"model": "deepseek-reasoner",
|
| 34 |
-
"
|
| 35 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 36 |
}
|
| 37 |
|
| 38 |
response = requests.post(url, json=data, headers=headers)
|
| 39 |
-
print(response.json()["choices"][0]["
|
| 40 |
|
| 41 |
"""
|
| 42 |
prompt: 一个汉字具有左右结构,左边是木,右边是乞,这个字是什么字,只需要回答这个字即可。
|
| 43 |
|
| 44 |
-
|
|
|
|
| 45 |
|
| 46 |
-
|
| 47 |
|
| 48 |
-
|
| 49 |
-
”字旁的字如“杨”、“林”等,但带有“乞”作为右边部分的字较不常见。
|
| 50 |
|
| 51 |
-
|
| 52 |
-
|
| 53 |
|
| 54 |
-
|
| 55 |
-
|
| 56 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 57 |
--------------------------------------------------
|
| 58 |
|
| 59 |
prompt: 请用我给你的4个数字,通过加、减、乘、除、括号,组成一个运算,使得结果为24。注意:数字需要全部使用我提供的数字,4468
|
| 60 |
|
| 61 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 62 |
|
| 63 |
-
|
| 64 |
-
算组合,然后检查结果是否正确。
|
| 65 |
|
| 66 |
-
|
| 67 |
-
似的。不过我这个组合并没有用全四个数字,因为用了两个4,可能是不是正确?
|
| 68 |
|
| 69 |
-
|
| 70 |
|
| 71 |
-
|
| 72 |
|
| 73 |
-
|
| 74 |
|
| 75 |
-
|
| 76 |
|
| 77 |
-
|
| 78 |
|
| 79 |
-
|
|
|
|
| 80 |
|
| 81 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 82 |
|
| 83 |
-
或者考虑减法结合乘法,比如,(8 -4) ×6 + 4×0?这显然不行
|
| 84 |
--------------------------------------------------
|
| 85 |
|
| 86 |
prompt: How many e in word deepseek?
|
| 87 |
-
|
| 88 |
-
Okay, so
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 89 |
|
| 90 |
1. D
|
| 91 |
2. E
|
|
@@ -96,30 +170,25 @@ Okay, so I need to figure out how many times the letter "e" appears in the word
|
|
| 96 |
7. E
|
| 97 |
8. K
|
| 98 |
|
| 99 |
-
|
| 100 |
|
| 101 |
-
|
| 102 |
-
Position 2: E – that's an e.
|
| 103 |
-
Position 3: E – another e.
|
| 104 |
-
Position 4: P – not an e.
|
| 105 |
-
Position 5: S – not an e.
|
| 106 |
-
Position 6: E – another e.
|
| 107 |
-
Position 7: E – another e.
|
| 108 |
-
Position 8: K – not an e.
|
| 109 |
|
| 110 |
-
|
| 111 |
</think>
|
| 112 |
|
| 113 |
-
The word "deepseek"
|
| 114 |
-
|
| 115 |
-
- **D**: Not an "e"
|
| 116 |
-
# consider s3
|
| 117 |
-
|
| 118 |
-
Alibabacloud.com offers a wide variety of articles about consider s3, easily find your consider s3 information here online.
|
| 119 |
|
| 120 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 121 |
|
| 122 |
-
|
| 123 |
|
| 124 |
"""
|
| 125 |
~~~
|
|
@@ -145,7 +214,7 @@ will update later
|
|
| 145 |
pip3 install lm-eval==0.4.8
|
| 146 |
|
| 147 |
```bash
|
| 148 |
-
|
| 149 |
```
|
| 150 |
| Metric | FP8 | INT4(BF16) |
|
| 151 |
| :------------------------ | :---------------------- | :--------------- |
|
|
|
|
| 27 |
~~~python
|
| 28 |
import requests
|
| 29 |
|
| 30 |
+
url = "http://localhost:12345/v1/chat/completions"
|
| 31 |
headers = {"Content-Type": "application/json"}
|
| 32 |
+
prompt="一个汉字具有左右结构,左边是木,右边是乞,这个字是什么字,只需要回答这个字即可。"
|
| 33 |
data = {
|
| 34 |
"model": "deepseek-reasoner",
|
| 35 |
+
"messages": [
|
| 36 |
+
{"role": "user", "content": prompt}
|
| 37 |
+
],
|
| 38 |
+
"max_tokens": 4096,
|
| 39 |
+
"top_p": 0.9,
|
| 40 |
+
"temperature": 0.6
|
| 41 |
}
|
| 42 |
|
| 43 |
response = requests.post(url, json=data, headers=headers)
|
| 44 |
+
print(response.json()["choices"][0]["message"]["content"])
|
| 45 |
|
| 46 |
"""
|
| 47 |
prompt: 一个汉字具有左右结构,左边是木,右边是乞,这个字是什么字,只需要回答这个字即可。
|
| 48 |
|
| 49 |
+
<think>
|
| 50 |
+
好的,我现在要解决的问题是:一个左右结构的汉字,左边是“木”,右边是“乞”,这个字是什么?首先,我需要确认这个字是否存在,然后找到它的正确读音和意思。
|
| 51 |
|
| 52 |
+
首先,我回忆一下常见的以“木”为偏旁的字,比如“林”、“松”、“柏”等等,但右边是“乞”的可能不太常见。接下来,我可以考虑拆解右边的“乞”字,看看有没有可能和其他偏旁组合成另一个字。
|
| 53 |
|
| 54 |
+
“乞”这个部首通常读作qǐ,比如“吃”、“迄”等字。左边是“木”,右边是“乞”,可能的结构是左右结构,左边是木字旁,右边是“乞”。那么这样的组合是否存在呢?
|
|
|
|
| 55 |
|
| 56 |
+
可能的方法包括查阅字典或者使用输入法尝试拼写。比如,用拼音输入法,输入muqi,看看有没有候选字。不过可能因为不常见,所以不一定能直接打出来。或者考虑五笔输入法,木是S,乞是TNB,
|
| 57 |
+
具体的五笔编码可能需要更详细的知识。
|
| 58 |
|
| 59 |
+
另外,我也可以考虑这个字是否有可能是某个生僻字或者古字。例如,在《说文解字》或者其他古代字典中是否有记载。比如,“杞”字是木字旁右边是“己”,而不是“乞”,所以可能和这个类似,但结
|
| 60 |
+
不同。
|
| 61 |
+
|
| 62 |
+
也有可能用户将右边的部首写错了,比如本来是“己”变成“杞”,但用户明确说是“乞”,所以需要确认是否存在这样的字。
|
| 63 |
+
|
| 64 |
+
再查一下在线字典,比如汉典或者新华字典,搜索木字旁加乞,看看是否有结果。或者使用部首检索,先找木部,再找右边三画的部首,因为“乞”是三画,然后看看对应的字。
|
| 65 |
+
|
| 66 |
+
经过查找,我发现确实有一个字是“杚”,由木和乞组成,左右结构。这个字的读音是gū,古同“榖”,但也有读音为gài,意思是平量。另外,还有一种可能读作qì,但需要进一步确认。
|
| 67 |
+
|
| 68 |
+
不过也有可能这个字并不常用,或者属于异体字,所以需要确认正确的答案。根据一些资料,"杚"确实存在,但可能比较生僻,通常在现代汉语中很少使用。因此,正确的答案应该是“杚”字。
|
| 69 |
+
</think>
|
| 70 |
+
|
| 71 |
+
杚
|
| 72 |
--------------------------------------------------
|
| 73 |
|
| 74 |
prompt: 请用我给你的4个数字,通过加、减、乘、除、括号,组成一个运算,使得结果为24。注意:数字需要全部使用我提供的数字,4468
|
| 75 |
|
| 76 |
+
<think> 嗯,用户给了我四个数字,4、4、6、8,需要组合起来用加减乘除和括号得到24。首先,我得回忆一下24点游戏的常见方法,通常可能需要用到乘法或者除法来增大或减小数值。用户可能已经尝试过
|
| 77 |
+
一些组合但没成功,所以需要仔细分析。
|
| 78 |
+
首先,这四个数字中有两个4,一个6和一个8。重复的数字可能会带来不同的组合方式。让我先考虑可能的乘法组合,因为24的因数比较多,比如4×6=24,3×8=24,或者12×2=24。但这里没有3,但可能
|
| 79 |
+
有其他组合方式。
|
| 80 |
+
|
| 81 |
+
比如,如果能用8×3的话,但怎么得到3呢?或者用6×4=24,但剩下的4和8需要处理。可能需要将剩下的数字通过运算抵消掉,比如加减。例如,如果有一个4×6=24,剩下的4和8可能需要组合成0或者1
|
| 82 |
+
,但这样可能不太行。
|
| 83 |
+
|
| 84 |
+
另一个思路是用减法或除法。比如,8/(某个数)得到3,然后乘以剩下的数。比如,8/( (6-4)/4 ),这样算的话,6-4=2,2/4=0.5,8/0.5=16,不够。或者反过来,可能的话。
|
| 85 |
+
|
| 86 |
+
或者考虑用括号改变运算顺序。比如,4 × (6 - (8 / 4)),这里8/4=2,6-2=4,4×4=16,还是不够。
|
| 87 |
+
|
| 88 |
+
再试试其他组合,比如4×4=16,然后6+8=14,16+14=30,也不对。或者8-6=2,4×4=16,16×2=32,也不对。
|
| 89 |
+
|
| 90 |
+
或者考虑用分数,比如8 × (4 - (6/4)),这里6/4=1.5,4-1.5=2.5,8×2.5=20,还是不够。
|
| 91 |
+
|
| 92 |
+
或者试试加减乘除混合。比如,(8 × (4 + 4)) - 6,这样是8×8=64,64-6=58,太大了。
|
| 93 |
+
|
| 94 |
+
或者,4 + 4 + 6 + 8=22,接近但不够。
|
| 95 |
+
|
| 96 |
+
可能需要用到除法来调整。比如,6 × (8 / (4 / 4)),这里4/4=1,8/1=8,6×8=48,不对。
|
| 97 |
+
|
| 98 |
+
或者,8 × 6 × (4 / 4) =48,还是不对。
|
| 99 |
+
|
| 100 |
+
或者,(8 - 6) × (4 × 4)=2×16=32,也不对。
|
| 101 |
+
|
| 102 |
+
这时候可能需要更复杂的组合,比如,考虑用括号来调整运算顺序。例如,4 × (4 × (8 - 6)),即4×4×2=32,还是不对。
|
| 103 |
+
|
| 104 |
+
或者,有没有可能用减法来调整?比如,8 × (6 - (4 / 4)),这里4/4=1,6-1=5,8×5=40,也不对。
|
| 105 |
+
|
| 106 |
+
再试试另一个角度,比如,8 × 3 =24,那如何得到3?比如,(6 - (8 / 4)),这里8/4=2,6-2=4,所以8×4=32,还是不行。
|
| 107 |
|
| 108 |
+
或者,(4 × (8 - 6)) × 4=4×2×4=32,还是不行。
|
|
|
|
| 109 |
|
| 110 |
+
或者,用除法来得到分数,比如6/( (8-4)/4 )=6/(4/4)=6/1=6,然后加上另一个4的话,得到10,也不够。
|
|
|
|
| 111 |
|
| 112 |
+
这时候可能需要重新排列组合,比如,4 × 6=24,然后剩下的4和8如何组合成0,或者1。例如,4 × 6 × (8 - 8),但这里没有重复的8,所以不行。或者用4 - (8/4)=4-2=2,然后6×2=12,也不对。
|
| 113 |
|
| 114 |
+
或者,试试用所有的四个数,比如,8 × (6 - (4 ÷ 4)),这里4÷4=1,6-1=5,8×5=40,还是不行。
|
| 115 |
|
| 116 |
+
或者,有没有可能用分数的方式?例如,(4 × 6) × (8 ÷ 4) ÷ 2,但这样用到了额外的数字2,不符合条件。
|
| 117 |
|
| 118 |
+
或者,4 + 4 + 6 + 8=22,不行。那可能需要更复杂的组合。
|
| 119 |
|
| 120 |
+
这时候可能需要更仔细地检查所有可能的组合。例如,考虑将���个4组合起来,比如4 + 4=8,然后8 × (8 -6)=8×2=16,还是不够。
|
| 121 |
|
| 122 |
+
或者,8 × 6 ÷ (4 - (4 ÷ something))。比如,假设是8×6/(4 - (4/ something)),需要让分母为2,这样得到48/2=24。那如何让分母为2?4 - (4/ something)=2,即4/ something=2,所以somethi
|
| 123 |
+
ng=2。那如何得到2?可能用6-4=2,但这样的话,分母是4 - (4/(6-4))=4 - (4/2)=4-2=2,所以整个式子是8×6/(4 - (4/(6-4)))=48/2=24。这样是否可行?
|
| 124 |
|
| 125 |
+
让我验证一下:6-4=2,4/2=2,4-2=2,8×6=48,48/2=24。是的,这样就满足了所有数字的使用:4、4、6、8。所以最终的表达式是8×6/(4 - (4/(6-4)))。或者用括号更清晰一点:8×6/((4 - (4/(6-
|
| 126 |
+
4))))
|
| 127 |
+
|
| 128 |
+
或者可以写成:(8 × 6) ÷ (4 - (4 ÷ (6 - 4))) =24。
|
| 129 |
+
|
| 130 |
+
这样确实用到了所有四个数字,并且通过括号调整了运算顺序。可能这个解法需要一定的技巧,特别是分式中的分母处理。用户可能没有想到这种分步运算,尤其是分母中的减法部分。
|
| 131 |
+
|
| 132 |
+
总结一下,正确的表达式是8×6/(4 - 4/(6-4)),也就是8×6/(4 - 4/2)=8×6/2=24。所以答案应该是这个。
|
| 133 |
+
</think>
|
| 134 |
+
|
| 135 |
+
我们可以通过以下运算得到24:
|
| 136 |
+
|
| 137 |
+
\[
|
| 138 |
+
\frac{8 \times 6}{4 - \frac{4}{6 - 4}} = 24
|
| 139 |
+
\]
|
| 140 |
+
|
| 141 |
+
**步骤解释:**
|
| 142 |
+
1. 先计算括号内的 \(6 - 4 = 2\);
|
| 143 |
+
2. 接着算 \(\frac{4}{2} = 2\);
|
| 144 |
+
3. 分母变为 \(4 - 2 = 2\);
|
| 145 |
+
4. 分子为 \(8 \times 6 = 48\);
|
| 146 |
+
5. 最后 \(48 \div 2 = 24\)。
|
| 147 |
+
|
| 148 |
+
这个运算用到了所有数字 4、4、6、8,并通过合理组合括号和分数达到了目标结果。
|
| 149 |
|
|
|
|
| 150 |
--------------------------------------------------
|
| 151 |
|
| 152 |
prompt: How many e in word deepseek?
|
| 153 |
+
<think>
|
| 154 |
+
Okay, so the user is asking how many times the letter "e" appears in the word "deepseek". Let me break this down step by step.
|
| 155 |
+
|
| 156 |
+
First, I need to make sure I spell the word correctly. The word is "deepseek". Let me write it out: D-E-E-P-S-E-E-K. Now, I'll go through each letter one by one and count the number of "e"s.
|
| 157 |
+
|
| 158 |
+
Starting with the first letter: D. That's not an E. The second letter is E. That's one. The third letter is another E, so that's two. The fourth letter is P, not an E. The fifth is S, not an E. The sixth is E, bringing the count to three. The seventh letter is another E, making it four. The eighth letter is K, which isn't an E.
|
| 159 |
+
|
| 160 |
+
Wait, let me check again to make sure I didn't miss any. D (1), E (2), E (3), P (4), S (5), E (6), E (7), K (8). So positions 2, 3, 6, and 7 are all E's. That's a total of four E's.
|
| 161 |
+
|
| 162 |
+
But let me verify once more. Maybe I miscounted. Let's list each letter with its position:
|
| 163 |
|
| 164 |
1. D
|
| 165 |
2. E
|
|
|
|
| 170 |
7. E
|
| 171 |
8. K
|
| 172 |
|
| 173 |
+
Yep, positions 2, 3, 6, 7 are E's. So that's four E's in total. Hmm, but wait, the word "deepseek" has two E's at the beginning, then after P and S, two more E's. So that's 2 + 2 = 4. I think that's correct.
|
| 174 |
|
| 175 |
+
Alternatively, maybe I can write the word out and underline each E: D E E P S E E K. Each underlined E is a count. Let's count them: 1, 2, 3, 4. Yes, four times.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 176 |
|
| 177 |
+
I don't think there are any other E's. The user might be confused if they think the last letter is E, but it's K. So the answer should be 4.
|
| 178 |
</think>
|
| 179 |
|
| 180 |
+
The word "deepseek" contains the letter **e** four times. Here's the breakdown:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 181 |
|
| 182 |
+
1. D
|
| 183 |
+
2. **E**
|
| 184 |
+
3. **E**
|
| 185 |
+
4. P
|
| 186 |
+
5. S
|
| 187 |
+
6. **E**
|
| 188 |
+
7. **E**
|
| 189 |
+
8. K
|
| 190 |
|
| 191 |
+
**Total:** 4 e's.
|
| 192 |
|
| 193 |
"""
|
| 194 |
~~~
|
|
|
|
| 214 |
pip3 install lm-eval==0.4.8
|
| 215 |
|
| 216 |
```bash
|
| 217 |
+
TORCH_DISTRIBUTED_DEBUG=INFO python -m lm_eval --model vllm --model_args "pretrained=OPEA/DeepSeek-R1-int4-AutoRound-awq-asym,tensor_parallel_size=8,dtype=bfloat16,max_model_len=65536,max_num_batched_tokens=65536,served_model_name=deepseek-reasoner" --batch_size 1 --device 'cuda'--trust_remote_code --tasks lambada_openai,hellaswag,piqa,winogrande,truthfulqa_mc1,boolq,arc_easy,arc_challenge,mmlu,openbookqa
|
| 218 |
```
|
| 219 |
| Metric | FP8 | INT4(BF16) |
|
| 220 |
| :------------------------ | :---------------------- | :--------------- |
|