A flexible data transformation tool for ML training formats (SFT, RLHF, Pretrain)
Project description
dtflow
简洁的数据格式转换工具,专为机器学习训练数据设计。
安装
pip install dtflow
# 可选依赖
pip install tiktoken # Token 统计(OpenAI 模型)
pip install transformers # Token 统计(HuggingFace 模型)
pip install datasets # HuggingFace Dataset 转换
快速开始
from dtflow import DataTransformer
# 加载数据
dt = DataTransformer.load("data.jsonl")
# 链式操作:过滤 -> 转换 -> 保存
(dt.filter(lambda x: x.score > 0.8)
.to(lambda x: {"q": x.question, "a": x.answer})
.save("output.jsonl"))
核心功能
数据加载与保存
# 支持 JSONL、JSON、CSV、Parquet
dt = DataTransformer.load("data.jsonl")
dt.save("output.jsonl")
# 从列表创建
dt = DataTransformer([{"q": "问题", "a": "答案"}])
数据过滤
# Lambda 过滤
dt.filter(lambda x: x.score > 0.8)
# 支持属性访问
dt.filter(lambda x: x.language == "zh")
数据转换
# 自定义转换
dt.to(lambda x: {"question": x.q, "answer": x.a})
# 使用预设模板
dt.to(preset="openai_chat", user_field="q", assistant_field="a")
预设模板
| 预设名称 | 输出格式 |
|---|---|
openai_chat |
{"messages": [{"role": "user", ...}, {"role": "assistant", ...}]} |
alpaca |
{"instruction": ..., "input": ..., "output": ...} |
sharegpt |
{"conversations": [{"from": "human", ...}, {"from": "gpt", ...}]} |
dpo_pair |
{"prompt": ..., "chosen": ..., "rejected": ...} |
simple_qa |
{"question": ..., "answer": ...} |
Token 统计
from dtflow import count_tokens, token_counter, token_filter, token_stats
# 计算 token 数量
count = count_tokens("Hello world", model="gpt-4")
# 添加 token_count 字段
dt.transform(token_counter("text")).save("with_tokens.jsonl")
# 按 token 长度过滤
dt.filter(token_filter("text", max_tokens=2048))
dt.filter(token_filter(["question", "answer"], min_tokens=10, max_tokens=4096))
# 统计 token 分布
stats = token_stats(dt.data, "text")
# {"total_tokens": 12345, "avg_tokens": 123, "min_tokens": 5, "max_tokens": 500, ...}
支持 tiktoken(OpenAI,默认)和 transformers 后端,自动检测:
# OpenAI 模型 -> 自动使用 tiktoken
count_tokens("Hello", model="gpt-4")
# HuggingFace/本地模型 -> 自动使用 transformers
count_tokens("Hello", model="Qwen/Qwen2-7B")
count_tokens("Hello", model="/home/models/qwen")
Messages Token 统计
专为多轮对话设计的 token 统计功能:
from dtflow import messages_token_counter, messages_token_filter, messages_token_stats
# 为每条数据添加 token 统计
dt.transform(messages_token_counter(model="gpt-4")) # 简单模式,输出总数
dt.transform(messages_token_counter(model="gpt-4", detailed=True)) # 详细模式
# 详细模式输出: {"total": 500, "user": 200, "assistant": 280, "system": 20, "turns": 5, ...}
# 按 token 数和轮数过滤
dt.filter(messages_token_filter(min_tokens=100, max_tokens=4096))
dt.filter(messages_token_filter(min_turns=2, max_turns=10))
# 统计整个数据集
stats = messages_token_stats(dt.data, model="gpt-4")
# {"count": 1000, "total_tokens": 500000, "user_tokens": 200000, "assistant_tokens": 290000, ...}
格式转换器
from dtflow import (
to_hf_dataset, from_hf_dataset, # HuggingFace Dataset
to_openai_batch, from_openai_batch, # OpenAI Batch API
to_llama_factory, # LLaMA-Factory Alpaca 格式
to_axolotl, # Axolotl 格式
messages_to_text, # messages 转纯文本
)
# HuggingFace Dataset 互转
ds = to_hf_dataset(dt.data)
ds.push_to_hub("my-dataset")
data = from_hf_dataset("tatsu-lab/alpaca", split="train")
# OpenAI Batch API
batch_input = dt.to(to_openai_batch(model="gpt-4o"))
results = from_openai_batch(batch_output)
# messages 转纯文本(支持 chatml/llama2/simple 模板)
dt.transform(messages_to_text(template="chatml"))
LLaMA-Factory 格式
完整支持 LLaMA-Factory 的 SFT 训练格式:
from dtflow import (
to_llama_factory, # Alpaca 格式(单轮)
to_llama_factory_sharegpt, # ShareGPT 格式(多轮对话)
to_llama_factory_vlm, # VLM Alpaca 格式
to_llama_factory_vlm_sharegpt, # VLM ShareGPT 格式
)
# Alpaca 格式
dt.transform(to_llama_factory()).save("alpaca.jsonl")
# 输出: {"instruction": "...", "input": "", "output": "..."}
# ShareGPT 格式(多轮对话)
dt.transform(to_llama_factory_sharegpt()).save("sharegpt.jsonl")
# 输出: {"conversations": [{"from": "human", "value": "..."}, {"from": "gpt", "value": "..."}], "system": "..."}
# VLM 格式(图片/视频)
dt.transform(to_llama_factory_vlm(images_field="images")).save("vlm.jsonl")
# 输出: {"instruction": "...", "output": "...", "images": ["/path/to/img.jpg"]}
dt.transform(to_llama_factory_vlm_sharegpt(images_field="images", videos_field="videos"))
# 输出: {"conversations": [...], "images": [...], "videos": [...]}
ms-swift 格式
支持 ModelScope ms-swift 的训练格式:
from dtflow import (
to_swift_messages, # 标准 messages 格式
to_swift_query_response, # query-response 格式
to_swift_vlm, # VLM 格式
)
# messages 格式
dt.transform(to_swift_messages()).save("swift_messages.jsonl")
# 输出: {"messages": [{"role": "user", "content": "..."}, {"role": "assistant", "content": "..."}]}
# query-response 格式(自动提取 history)
dt.transform(to_swift_query_response(query_field="messages")).save("swift_qr.jsonl")
# 输出: {"query": "...", "response": "...", "system": "...", "history": [["q1", "a1"], ...]}
# VLM 格式
dt.transform(to_swift_vlm(images_field="images")).save("swift_vlm.jsonl")
# 输出: {"messages": [...], "images": ["/path/to/img.jpg"]}
其他操作
# 采样
dt.sample(100) # 随机采样 100 条
dt.head(10) # 前 10 条
dt.tail(10) # 后 10 条
# 分割
train, test = dt.split(ratio=0.8, shuffle=True, seed=42)
# 统计
stats = dt.stats() # 总数、字段信息
count = dt.count(lambda x: x.score > 0.9)
# 打乱
dt.shuffle(seed=42)
CLI 命令
# 数据采样
dt sample data.jsonl --num=10
dt sample data.csv --num=100 --sample_type=head
# 数据转换 - 预设模式
dt transform data.jsonl --preset=openai_chat
dt transform data.jsonl --preset=alpaca
# 数据转换 - 配置文件模式
dt transform data.jsonl # 首次运行生成配置文件
# 编辑 .dt/data.py 后再次运行
dt transform data.jsonl --num=100 # 执行转换
# 数据清洗
dt clean data.jsonl --drop-empty # 删除任意空值记录
dt clean data.jsonl --drop-empty=text,answer # 删除指定字段为空的记录
dt clean data.jsonl --min-len=text:10 # text 字段最少 10 字符
dt clean data.jsonl --max-len=text:1000 # text 字段最多 1000 字符
dt clean data.jsonl --keep=question,answer # 只保留这些字段
dt clean data.jsonl --drop=metadata # 删除指定字段
dt clean data.jsonl --strip # 去除字符串首尾空白
dt clean data.jsonl --strip --drop-empty=text --min-len=text:10 -o clean.jsonl # 组合使用
# 数据去重
dt dedupe data.jsonl # 全量精确去重
dt dedupe data.jsonl --key=text # 按字段精确去重
dt dedupe data.jsonl --key=text --similar=0.8 # 相似度去重
# 文件拼接
dt concat a.jsonl b.jsonl -o merged.jsonl
# 数据统计
dt stats data.jsonl
错误处理
# 跳过错误项(默认)
dt.to(transform_func, on_error="skip")
# 抛出异常
dt.to(transform_func, on_error="raise")
# 保留原始数据
dt.to(transform_func, on_error="keep")
# 返回错误信息
result, errors = dt.to(transform_func, return_errors=True)
设计哲学
函数式优于类继承
不需要复杂的 OOP 抽象,直接用函数解决问题:
# ✅ 简单直接
dt.to(lambda x: {"q": x.question, "a": x.answer})
# ❌ 不需要这种设计
class MyFormatter(BaseFormatter):
def format(self, item): ...
预设是便利层,不是核心抽象
90% 的需求用 transform(lambda x: ...) 就能解决。预设只是常见场景的快捷方式:
# 预设:常见场景的便利函数
dt.to(preset="openai_chat")
# 自定义:完全控制转换逻辑
dt.to(lambda x: {
"messages": [
{"role": "user", "content": x.q},
{"role": "assistant", "content": x.a}
]
})
KISS 原则
- 一个核心类
DataTransformer搞定所有操作 - 链式 API,代码像自然语言
- 属性访问
x.field代替x["field"] - 不过度设计,不追求"可扩展框架"
实用主义
不追求学术上的完美抽象,只提供足够好用的工具。
License
MIT
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file dtflow-0.3.0.tar.gz.
File metadata
- Download URL: dtflow-0.3.0.tar.gz
- Upload date:
- Size: 58.6 kB
- Tags: Source
- Uploaded using Trusted Publishing? Yes
- Uploaded via: twine/6.1.0 CPython/3.13.7
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
8a60922abea2f05902b277eea6dd5bca2c91c980d975d30f832faf51f79620d0
|
|
| MD5 |
259d45280eb956fc632a34d5a2149364
|
|
| BLAKE2b-256 |
e14e8594e63bee1a4a0ae34fe547459a7061378c152afe353c28e63447cd4262
|
Provenance
The following attestation bundles were made for dtflow-0.3.0.tar.gz:
Publisher:
python-publish.yml on ai-ivan/data_transformer
-
Statement:
-
Statement type:
https://in-toto.io/Statement/v1 -
Predicate type:
https://docs.pypi.org/attestations/publish/v1 -
Subject name:
dtflow-0.3.0.tar.gz -
Subject digest:
8a60922abea2f05902b277eea6dd5bca2c91c980d975d30f832faf51f79620d0 - Sigstore transparency entry: 768800664
- Sigstore integration time:
-
Permalink:
ai-ivan/data_transformer@72060d766d7485754b52150eb4fef757f195bf2d -
Branch / Tag:
refs/tags/v0.3.0 - Owner: https://github.com/ai-ivan
-
Access:
private
-
Token Issuer:
https://token.actions.githubusercontent.com -
Runner Environment:
github-hosted -
Publication workflow:
python-publish.yml@72060d766d7485754b52150eb4fef757f195bf2d -
Trigger Event:
push
-
Statement type:
File details
Details for the file dtflow-0.3.0-py3-none-any.whl.
File metadata
- Download URL: dtflow-0.3.0-py3-none-any.whl
- Upload date:
- Size: 53.8 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? Yes
- Uploaded via: twine/6.1.0 CPython/3.13.7
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
419a327a30e56218bce8a5c784850102420556b7f7068a2a5999ee0cecabcd25
|
|
| MD5 |
403bb163e1e89e0397e6ca101b36ee26
|
|
| BLAKE2b-256 |
9c140996da0111dbcb601d0959112ff87f18d8d68332b6cb277936554ee81511
|
Provenance
The following attestation bundles were made for dtflow-0.3.0-py3-none-any.whl:
Publisher:
python-publish.yml on ai-ivan/data_transformer
-
Statement:
-
Statement type:
https://in-toto.io/Statement/v1 -
Predicate type:
https://docs.pypi.org/attestations/publish/v1 -
Subject name:
dtflow-0.3.0-py3-none-any.whl -
Subject digest:
419a327a30e56218bce8a5c784850102420556b7f7068a2a5999ee0cecabcd25 - Sigstore transparency entry: 768800667
- Sigstore integration time:
-
Permalink:
ai-ivan/data_transformer@72060d766d7485754b52150eb4fef757f195bf2d -
Branch / Tag:
refs/tags/v0.3.0 - Owner: https://github.com/ai-ivan
-
Access:
private
-
Token Issuer:
https://token.actions.githubusercontent.com -
Runner Environment:
github-hosted -
Publication workflow:
python-publish.yml@72060d766d7485754b52150eb4fef757f195bf2d -
Trigger Event:
push
-
Statement type: