conversations listlengths 2 200 | source stringclasses 4 values |
|---|---|
[
{
"from": "human",
"value": "Pull stock data from this API 'https://www.alphavantage.co/query?function=TIME_SERIES_DAILY&symbol=MSFT&apikey=demo', and visualize the opening, closing, high, and low stock prices over time using Matplotlib. The visualization must include the volume of shares traded each day."
... | perfectblend |
[
{
"from": "human",
"value": "Write Python code to solve the task.\nYou are given string s. Your task is to determine if the given string s contains two non-overlapping substrings \"AB\" and \"BA\" (the substrings can go in any order).\n\nInput\n\nThe only line of input contains a string s of length between ... | perfectblend |
[
{
"from": "human",
"value": "Write Python code to solve the task.\nChef and Roma are playing a game. Rules of the game are quite simple.\nInitially there are N piles of stones on the table.\nIn each turn, a player can choose one pile and remove it from the table.\nEach player want to maximize the total numb... | perfectblend |
[
{
"from": "human",
"value": "What would be the easiest way to separate the directory name from the file name when dealing with `SaveFileDialog.FileName` in C#?"
},
{
"from": "gpt",
"value": " The user is asking about the easiest way to separate the directory name (folder path) from the file name... | perfectblend |
[
{
"from": "human",
"value": "A school is planning a community outreach program. Each classroom must raise $200 for this activity. Classroom A has raised $20 from each of two families, $10 from each of eight families, and $5 from each of some families. They need $30 more to reach the goal. How many families ... | perfectblend |
[
{
"from": "human",
"value": "Can you summarize the experiences of Professor Pantelis and Anthony Stratford, and how they have used their personal experiences to inform their work in mental health?: Professor Pantelis, a professor of medicine at the University of Melbourne, grew up watching his beloved aunt ... | perfectblend |
[
{
"from": "human",
"value": "Describe a sequence of emails in a customer journey for new customers that registered at an online casino business but did not become paying customers yet. That purpose of this customer journey is to convert these customers from registration to paying. Your answer must contain a... | perfectblend |
[
{
"from": "human",
"value": "2 toucans are sitting on a tree limb. A certain number of toucans join them, and now there are 3 toucans in all. How many toucans joined them?"
},
{
"from": "gpt",
"value": " The user is asking a simple math word problem. Let me analyze it:\n\n - Initial number of to... | perfectblend |
[
{
"from": "human",
"value": "a and b together have rs. 1210. If 2/3 of a's amount is equal to a certain fraction of b's amount, and b has rs. 484, what is the fraction of b's amount that is equal to 2/3 of a's amount?"
},
{
"from": "gpt",
"value": " The user wants me to solve a math problem.\n\n... | perfectblend |
[
{
"from": "human",
"value": "Tool available:\n[1] Python interpreter\nWhen you send a message containing Python code to python, it will be executed in a stateful Jupyter notebook environment.\nSolve the following math problem step-by-step.\nLeo had 400 marbles in a jar. He packed the marbles with x marbles ... | perfectblend |
[
{
"from": "human",
"value": "How do I find the courage to buy a house?"
},
{
"from": "gpt",
"value": " The user is asking for advice on finding the courage to buy a house. This is a significant life decision that involves financial commitment, long-term planning, and emotional readiness. I shoul... | perfectblend |
[
{
"from": "human",
"value": "Translate these into traditional Chinese (not simplifies Chinese): Earth, the Solar System, Oort Cloud, Local Fluff, Local Bubble, Orion Arm, Milky Way Galaxy, Local Group, Virgo Supercluster, Laniakea Supercluster, Universe"
},
{
"from": "gpt",
"value": " The user w... | perfectblend |
[
{
"from": "human",
"value": "Tool available:\n[1] Python interpreter\nWhen you send a message containing Python code to python, it will be executed in a stateful Jupyter notebook environment.\nSolve the following math problem step-by-step.\nA library has 7 rectangular tables and the remaining tables will be... | perfectblend |
[
{
"from": "human",
"value": "If Layla scored x goals in four hockey games and Kristin scored 24 fewer goals in the same four games, calculate the average number of goals the two scored.\nIf we know the answer to the above question is 92, what is the value of unknown variable x?"
},
{
"from": "gpt",
... | perfectblend |
[
{
"from": "human",
"value": "How can we provide better healthcare in India?"
},
{
"from": "gpt",
"value": " The user is asking a broad question about improving healthcare in India. This is a complex, multi-faceted issue that requires a comprehensive answer covering various aspects of the healthc... | perfectblend |
[
{
"from": "human",
"value": "Tool available:\n[1] Python interpreter\nWhen you send a message containing Python code to python, it will be executed in a stateful Jupyter notebook environment.\nSolve the following math problem step-by-step.\nA party venue has 4 tables that seat 6 people each, 16 tables that ... | perfectblend |
[
{
"from": "human",
"value": "Write Python code to solve the task.\nBased on the information of the time when study started and the time when study ended, check whether the total time studied in one day is t or more, and if not, create a program to find the shortage time. Time is one unit per hour, and minut... | perfectblend |
[
{
"from": "human",
"value": "The average speed of a car is 1 4/5 times the average speed of a bike. A tractor covers 575 km in 23 hours. The speed of the bike is some times the speed of the tractor. The car covers 630 km in 7 hours. What is the ratio of the speed of the bike to the speed of the tractor?"
... | perfectblend |
[
{
"from": "human",
"value": "After improving her speed in double dutch, Jan can now do twice as many skips per minute compared to her previous speed of 70 skips. How many skips does she do in a span of 5 minutes?"
},
{
"from": "gpt",
"value": " The user wants me to solve a math word problem.\n\n... | perfectblend |
[
{
"from": "human",
"value": "When did Emperor Gaozong of the Tang Dynasty built the church to collect Buddhist materials?"
},
{
"from": "gpt",
"value": " The user is asking about Emperor Gaozong of the Tang Dynasty building a church to collect Buddhist materials. This seems to be a mistranslatio... | perfectblend |
[
{
"from": "human",
"value": "TASK DEFINITION: In this task, you are given reviews written about the restaurants in Bengali. You are expected to classify the sentiment of the reviews into two classes: positive or negative.\nPROBLEM: এই শহরে সেরা BBQ,গ্রিলস ... তাদের তাদের সেহরি / ইফতারের খাবার পছন্দনীয় ... ত... | perfectblend |
[
{
"from": "human",
"value": "Your class is having a pizza party. You buy 5 pizzas. Each pizza has 4 slices. How many slices are there altogether?"
},
{
"from": "gpt",
"value": " The user is asking a simple math problem. Let's break it down:\n\n1. Number of pizzas: 5\n2. Slices per pizza: 4\n3. T... | perfectblend |
[
{
"from": "human",
"value": "What are the benefits of promoting diversity and inclusion in leadership positions within a company?"
},
{
"from": "gpt",
"value": " The user is asking about the benefits of promoting diversity and inclusion (D&I) in leadership positions within a company. This is a b... | perfectblend |
[
{
"from": "human",
"value": "In a 1000-word essay, critically analyze the effects of automation on job satisfaction using at least three empirical research studies as evidence. Your analysis should cover both positive and negative impacts on job satisfaction, and consider factors such as job security, auton... | perfectblend |
[
{
"from": "human",
"value": "Solve the following math problem step-by-step.\nSimplify your answer as much as possible. Present your final answer as \\boxed{Your Answer}.\nThe double-bar graph shows the number of home runs hit by McGwire and Sosa during each month of the 1998 baseball season. At the end of w... | perfectblend |
[
{
"from": "human",
"value": "Jessica is making an apple pie. She knows that each serving requires 1.5 apples and she has x quests. She plans to make 3 pies, which each contain 8 servings. If her guests finish all the pie, on average , Each guest eats 3 apples. What is the value of unknown variable x? What i... | perfectblend |
[
{
"from": "human",
"value": "What is the total cost of buying a wallet for $22 and a purse that is priced $3 less than four times the cost of the wallet?"
},
{
"from": "gpt",
"value": " The user wants me to calculate the total cost of buying a wallet and a purse.\n\nGiven:\n- Wallet costs $22\n-... | perfectblend |
[
{
"from": "human",
"value": "What is X \\div 0. 004 is equal to 25. What is the value of unknown variable X?"
},
{
"from": "gpt",
"value": " The user is asking: \"What is X \\div 0.004 is equal to 25. What is the value of unknown variable X?\"\n\n This is an algebra problem. The equation is:\n X... | perfectblend |
[
{
"from": "human",
"value": "If Emma was able to make 10 pieces of toast bread and had 1 slice of bread left after her little cousin Andy ate 3 slices at two different times, and each piece of toast bread requires 2 slices, how many slices were in the original loaf of bread?"
},
{
"from": "gpt",
... | perfectblend |
[
{
"from": "human",
"value": "1 of 1 In a large bowl, mix together the flour, baking powder, baking soda and salt.\n1 of 3 Spoon flour mixture into bottom of jar and press down hard to compact (a small spice jar works well for compacting). It needs to be at about the 1-1/4-cup mark on the jar to insure that ... | perfectblend |
[
{
"from": "human",
"value": "please write a powerful tweet (use lists an emojis to emphasize the points) that problem always starts with definitions: semantically HealthCare is care about health, healthy, by logic it means exercise, healthy diet, health sleep, healthy relationships, measures to improve heal... | perfectblend |
End of preview. Expand
in Data Studio
Kimi-K2.5 Eagle3 Training Data
This dataset contains the instruction-following data used to train an Eagle3 MTP draft model for Kimi-K2.5 with TorchSpec.
All responses were regenerated by running Kimi-K2.5 via Engine rather than taken from the original datasets. This is critical for speculative decoding training: the draft model must learn the exact token-level distribution of the target model it is accelerating.
The trained Eagle3 draft model is available at lightseekorg/kimi-k2.5-eagle3. If you find this draft model useful, please give our project TorchSpec a 🌟 on GitHub.
Data source
Due to inference resource constraints, some source datasets are only partially regenerated. Here is the list of source datasets used in this mix:
| Dataset | Source | # Samples |
|---|---|---|
| mlabonne/open-perfectblend | perfectblend |
296,034 |
| liuhaotian/LLaVA-Instruct-150K | llava_instruct |
123,102 |
| HuggingFaceTB/smoltalk | smoltalk_cn |
48,333 |
| daviddjtzafon/continual-tool-kimi-k2.5 | continual_tool_kimi |
4,370 |
| crownelius/KimiK2.5-2000x-formatted | kimi_2000x |
2,144 |
| crownelius/Creative-Writing-KimiK2.5-Cleaned | creative_writing |
1,393 |
| DCAgent2/terminal_bench_2 | dcagent |
873 |
| crownelius/Creative-Writing-Reasoning-KimiK2.5-600x | creative_writing_reasoning |
655 |
| Total | 476,904 |
Data format
Each sample contains two fields:
conversations: a list of turns, each withfrom(human/gpt/system) andvalue(string).source: the name of the source dataset (see table above).
{
"conversations": [
{"from": "human", "value": "What is the capital of France?"},
{"from": "gpt", "value": "The capital of France is Paris."}
],
"source": "perfectblend"
}
Multimodal samples (llava_instruct) use OpenAI vision format in the value field — a list of image_url and text objects — with local image paths replaced by public COCO URLs (http://images.cocodataset.org/train2017/{filename}).
Function-call samples (continual_tool_kimi) use Kimi-K2.5's special token format for tool calls:
<|tool_calls_section_begin|><|tool_call_begin|>{id}<|tool_call_argument_begin|>{args_json}<|tool_call_end|><|tool_calls_section_end|>
Tool results are serialized as human turns with the prefix ## Return of {call_id}\n.
Training
See TorchSpec for the full training recipe, configuration, and evaluation results.
License
Apache 2.0. All source datasets are Apache 2.0 or MIT licensed.
- Downloads last month
- 28