Croissant: A Metadata Format for ML-Ready Datasets
Paper
β’
2403.19546
β’
Published
OpenSLR
non-profit
AI & ML interests
None defined yet.
Recent Activity
View all activity
openslr's activity
lhoestqΒ
authored
a
paper
3 days ago
Post
1846
Introducing ππ
π’π§πππππ‘: the best public math pre-training dataset with 50B+ tokens!
HuggingFaceTB/finemath
Math remains challenging for LLMs and by training on FineMath we see considerable gains over other math datasets, especially on GSM8K and MATH.
We build the dataset by:
π οΈ carefully extracting math data from Common Crawl;
π iteratively filtering and recalling high quality math pages using a classifier trained on synthetic annotations to identify math reasoning and deduction.
We conducted a series of ablations comparing the performance of Llama-3.2-3B-Base after continued pre-training on FineMath and observe notable gains compared to the baseline model and other public math datasets.
We hope this helps advance the performance of LLMs on math and reasoning! π
Weβre also releasing all the ablation models as well as the evaluation code.
HuggingFaceTB/finemath-6763fb8f71b6439b653482c2
HuggingFaceTB/finemath
Math remains challenging for LLMs and by training on FineMath we see considerable gains over other math datasets, especially on GSM8K and MATH.
We build the dataset by:
π οΈ carefully extracting math data from Common Crawl;
π iteratively filtering and recalling high quality math pages using a classifier trained on synthetic annotations to identify math reasoning and deduction.
We conducted a series of ablations comparing the performance of Llama-3.2-3B-Base after continued pre-training on FineMath and observe notable gains compared to the baseline model and other public math datasets.
We hope this helps advance the performance of LLMs on math and reasoning! π
Weβre also releasing all the ablation models as well as the evaluation code.
HuggingFaceTB/finemath-6763fb8f71b6439b653482c2
Post
1591
Made a HF Dataset editor a la gg sheets here:
lhoestq/dataset-spreadsheets
With Dataset Spreadsheets:
βοΈ Edit datasets in the UI
π Share link with collaborators
π Use locally in DuckDB or Python
Available for the 100,000+ parquet datasets on HF :)
With Dataset Spreadsheets:
βοΈ Edit datasets in the UI
π Share link with collaborators
π Use locally in DuckDB or Python
Available for the 100,000+ parquet datasets on HF :)
albertvillanovaΒ
posted
an
update
about 1 month ago
Post
1366
π¨ How green is your model? π± Introducing a new feature in the Comparator tool: Environmental Impact for responsible #LLM research!
π open-llm-leaderboard/comparator
Now, you can not only compare models by performance, but also by their environmental footprint!
π The Comparator calculates COβ emissions during evaluation and shows key model characteristics: evaluation score, number of parameters, architecture, precision, type... π οΈ
Make informed decisions about your model's impact on the planet and join the movement towards greener AI!
π open-llm-leaderboard/comparator
Now, you can not only compare models by performance, but also by their environmental footprint!
π The Comparator calculates COβ emissions during evaluation and shows key model characteristics: evaluation score, number of parameters, architecture, precision, type... π οΈ
Make informed decisions about your model's impact on the planet and join the movement towards greener AI!
albertvillanovaΒ
posted
an
update
about 2 months ago
Post
1459
π New feature of the Comparator of the π€ Open LLM Leaderboard: now compare models with their base versions & derivatives (finetunes, adapters, etc.). Perfect for tracking how adjustments affect performance & seeing innovations in action. Dive deeper into the leaderboard!
π οΈ Here's how to use it:
1. Select your model from the leaderboard.
2. Load its model tree.
3. Choose any base & derived models (adapters, finetunes, merges, quantizations) for comparison.
4. Press Load.
See side-by-side performance metrics instantly!
Ready to dive in? π Try the π€ Open LLM Leaderboard Comparator now! See how models stack up against their base versions and derivatives to understand fine-tuning and other adjustments. Easier model analysis for better insights! Check it out here: open-llm-leaderboard/comparator π
π οΈ Here's how to use it:
1. Select your model from the leaderboard.
2. Load its model tree.
3. Choose any base & derived models (adapters, finetunes, merges, quantizations) for comparison.
4. Press Load.
See side-by-side performance metrics instantly!
Ready to dive in? π Try the π€ Open LLM Leaderboard Comparator now! See how models stack up against their base versions and derivatives to understand fine-tuning and other adjustments. Easier model analysis for better insights! Check it out here: open-llm-leaderboard/comparator π
albertvillanovaΒ
posted
an
update
about 2 months ago
Post
3114
π Exciting update! You can now compare multiple models side-by-side with the Hugging Face Open LLM Comparator! π
open-llm-leaderboard/comparator
Dive into multi-model evaluations, pinpoint the best model for your needs, and explore insights across top open LLMs all in one place. Ready to level up your model comparison game?
open-llm-leaderboard/comparator
Dive into multi-model evaluations, pinpoint the best model for your needs, and explore insights across top open LLMs all in one place. Ready to level up your model comparison game?
albertvillanovaΒ
posted
an
update
about 2 months ago
Post
1220
π¨ Instruct-tuning impacts models differently across families! Qwen2.5-72B-Instruct excels on IFEval but struggles with MATH-Hard, while Llama-3.1-70B-Instruct avoids MATH performance loss! Why? Can they follow the format in examples? π Compare models:
open-llm-leaderboard/comparator
albertvillanovaΒ
posted
an
update
2 months ago
Post
1910
Finding the Best SmolLM for Your Project
Need an LLM assistant but unsure which hashtag#smolLM to run locally? With so many models available, how can you decide which one suits your needs best? π€
If the model youβre interested in is evaluated on the Hugging Face Open LLM Leaderboard, thereβs an easy way to compare them: use the model Comparator tool: open-llm-leaderboard/comparator
Letβs walk through an exampleπ
Letβs compare two solid options:
- Qwen2.5-1.5B-Instruct from Alibaba Cloud Qwen (1.5B params)
- gemma-2-2b-it from Google (2.5B params)
For an assistant, you want a model thatβs great at instruction following. So, how do these two models stack up on the IFEval task?
What about other evaluations?
Both models are close in performance on many other tasks, showing minimal differences. Surprisingly, the 1.5B Qwen model performs just as well as the 2.5B Gemma in many areas, even though it's smaller in size! π
This is a great example of how parameter size isnβt everything. With efficient design and training, a smaller model like Qwen2.5-1.5B can match or even surpass larger models in certain tasks.
Looking for other comparisons? Drop your model suggestions below! π
Need an LLM assistant but unsure which hashtag#smolLM to run locally? With so many models available, how can you decide which one suits your needs best? π€
If the model youβre interested in is evaluated on the Hugging Face Open LLM Leaderboard, thereβs an easy way to compare them: use the model Comparator tool: open-llm-leaderboard/comparator
Letβs walk through an exampleπ
Letβs compare two solid options:
- Qwen2.5-1.5B-Instruct from Alibaba Cloud Qwen (1.5B params)
- gemma-2-2b-it from Google (2.5B params)
For an assistant, you want a model thatβs great at instruction following. So, how do these two models stack up on the IFEval task?
What about other evaluations?
Both models are close in performance on many other tasks, showing minimal differences. Surprisingly, the 1.5B Qwen model performs just as well as the 2.5B Gemma in many areas, even though it's smaller in size! π
This is a great example of how parameter size isnβt everything. With efficient design and training, a smaller model like Qwen2.5-1.5B can match or even surpass larger models in certain tasks.
Looking for other comparisons? Drop your model suggestions below! π
albertvillanovaΒ
posted
an
update
2 months ago
Post
1946
π¨ Weβve just released a new tool to compare the performance of models in the π€ Open LLM Leaderboard: the Comparator π
open-llm-leaderboard/comparator
Want to see how two different versions of LLaMA stack up? Letβs walk through a step-by-step comparison of LLaMA-3.1 and LLaMA-3.2. π¦π§΅π
1/ Load the Models' Results
- Go to the π€ Open LLM Leaderboard Comparator: open-llm-leaderboard/comparator
- Search for "LLaMA-3.1" and "LLaMA-3.2" in the model dropdowns.
- Press the Load button. Ready to dive into the results!
2/ Compare Metric Results in the Results Tab π
- Head over to the Results tab.
- Here, youβll see the performance metrics for each model, beautifully color-coded using a gradient to highlight performance differences: greener is better! π
- Want to focus on a specific task? Use the Task filter to hone in on comparisons for tasks like BBH or MMLU-Pro.
3/ Check Config Alignment in the Configs Tab βοΈ
- To ensure youβre comparing apples to apples, head to the Configs tab.
- Review both modelsβ evaluation configurations, such as metrics, datasets, prompts, few-shot configs...
- If something looks off, itβs good to know before drawing conclusions! β
4/ Compare Predictions by Sample in the Details Tab π
- Curious about how each model responds to specific inputs? The Details tab is your go-to!
- Select a Task (e.g., MuSR) and then a Subtask (e.g., Murder Mystery) and then press the Load Details button.
- Check out the side-by-side predictions and dive into the nuances of each modelβs outputs.
5/ With this tool, itβs never been easier to explore how small changes between model versions affect performance on a wide range of tasks. Whether youβre a researcher or enthusiast, you can instantly visualize improvements and dive into detailed comparisons.
π Try the π€ Open LLM Leaderboard Comparator now and take your model evaluations to the next level!
open-llm-leaderboard/comparator
Want to see how two different versions of LLaMA stack up? Letβs walk through a step-by-step comparison of LLaMA-3.1 and LLaMA-3.2. π¦π§΅π
1/ Load the Models' Results
- Go to the π€ Open LLM Leaderboard Comparator: open-llm-leaderboard/comparator
- Search for "LLaMA-3.1" and "LLaMA-3.2" in the model dropdowns.
- Press the Load button. Ready to dive into the results!
2/ Compare Metric Results in the Results Tab π
- Head over to the Results tab.
- Here, youβll see the performance metrics for each model, beautifully color-coded using a gradient to highlight performance differences: greener is better! π
- Want to focus on a specific task? Use the Task filter to hone in on comparisons for tasks like BBH or MMLU-Pro.
3/ Check Config Alignment in the Configs Tab βοΈ
- To ensure youβre comparing apples to apples, head to the Configs tab.
- Review both modelsβ evaluation configurations, such as metrics, datasets, prompts, few-shot configs...
- If something looks off, itβs good to know before drawing conclusions! β
4/ Compare Predictions by Sample in the Details Tab π
- Curious about how each model responds to specific inputs? The Details tab is your go-to!
- Select a Task (e.g., MuSR) and then a Subtask (e.g., Murder Mystery) and then press the Load Details button.
- Check out the side-by-side predictions and dive into the nuances of each modelβs outputs.
5/ With this tool, itβs never been easier to explore how small changes between model versions affect performance on a wide range of tasks. Whether youβre a researcher or enthusiast, you can instantly visualize improvements and dive into detailed comparisons.
π Try the π€ Open LLM Leaderboard Comparator now and take your model evaluations to the next level!
albertvillanovaΒ
posted
an
update
3 months ago
Post
1524
Check out the new Structured #Wikipedia dataset by Wikimedia Enterprise: abstract, infobox, structured sections, main image,...
Currently in early beta (English & French). Explore it and give feedback: wikimedia/structured-wikipedia
More info: https://enterprise.wikimedia.com/blog/hugging-face-dataset/
@sdelbecque @resquito-wmf
Currently in early beta (English & French). Explore it and give feedback: wikimedia/structured-wikipedia
More info: https://enterprise.wikimedia.com/blog/hugging-face-dataset/
@sdelbecque @resquito-wmf
albertvillanovaΒ
updated
2
datasets
4 months ago
Post
4043
Hey ! I'm working on a 100% synthetic Dataset Hub here (you can search for any kind of datasets an the app invents them). The link is here:
infinite-dataset-hub/infinite-dataset-hub
Question for the Community:
Which models should I use to generate images and audio samples for those datasets ? π€
Question for the Community:
Which models should I use to generate images and audio samples for those datasets ? π€
-
- 4 replies
anton-lΒ
authored
a
paper
6 months ago
albertvillanovaΒ
posted
an
update
7 months ago
Post
2699
Easily convert your script-based datasets to Parquet and explore them in the dataset viewer. π
π οΈ Use @huggingface Datasets CLI:
$ ππππππππ-πππ πππππππ_ππ_πππππππ πππ΄ππ½π°πΌπ΄/π³π°ππ°ππ΄π_π½π°πΌπ΄
Learn more: https://huggingface.co/docs/datasets/main/en/cli#convert-to-parquet
#Data #AI
π οΈ Use @huggingface Datasets CLI:
$ ππππππππ-πππ πππππππ_ππ_πππππππ πππ΄ππ½π°πΌπ΄/π³π°ππ°ππ΄π_π½π°πΌπ΄
Learn more: https://huggingface.co/docs/datasets/main/en/cli#convert-to-parquet
#Data #AI
albertvillanovaΒ
posted
an
update
8 months ago
Post
4059
Recently, the Hugging Face π€ datasets team met with the Language Technologies team led by Marta Villegas (
@mvillegas
) at Barcelona Supercomputing Center
@BSC-LT
. Eager to collaborate to promote AI across Catalan, Spanish, Basque, and Galician languages and share open-source datasets/models. π€ #AI #LanguageTech #OpenSource
- 1 reply
albertvillanovaΒ
posted
an
update
8 months ago
Post
1661
π We recently released datasets 2.19.0! π¦
π₯ What's New:
- Polars integration π»ββοΈ
- fsspec support for conversion to JSON, CSV, and Parquet
- Mode parameter for Image feature
- CLI function to convert script-datasets to Parquet
- Dataset.take and Dataset.skip
Plus, a bunch of general improvements & bug fixes!
Check out the release notes: https://github.com/huggingface/datasets/releases/tag/2.19.0
Upgrade now and power up your data workflows! π₯
π₯ What's New:
- Polars integration π»ββοΈ
- fsspec support for conversion to JSON, CSV, and Parquet
- Mode parameter for Image feature
- CLI function to convert script-datasets to Parquet
- Dataset.take and Dataset.skip
Plus, a bunch of general improvements & bug fixes!
Check out the release notes: https://github.com/huggingface/datasets/releases/tag/2.19.0
Upgrade now and power up your data workflows! π₯
- 2 replies
Post
3058
β¨ Easy Synthetic Dataset File Generation using LLM DataGen ! Link: https://huggingface.co/spaces/lhoestq/LLM_DataGen
features + how it works:
βοΈ Generate the dataset content you want just by entering a file name
π‘ Optionally specify the column names you need
π¨ The dataset is streamed and generated on-the-fly in JSON Lines format
β Generation is constrained to always output valid JSON
How does this work ?
1/ Enter a file name
2/ The model generates column names for such a file. Using structured generation, it can generate 2 to 5 column names using lower case characters and underscores. I use a prompt that asks to generate column names for a realistic dataset and low temperature.
3/ The columns are used to update the Finite State Machine for the dataset content structured generation, so that it is used to generate JSON objects using those columns
4/ The model generates JSON objects using structured generation again, using the updated Finite State Machine. I use a prompt that asks for realistic data and a temperature of 1.
> Why update a Finite State Machine instead of re-creating one ?
Creating one can take up to 30sec, while updating one takes 0.1s (though it requires to manipulate a graph which is not easy to implement)
> Batched generation is faster, why not use it ?
Generate in batches is faster but tends to generate duplicates for this demo.
Further work can be to provide different prompts (one per sequence in the batch) to end up with a different distribution of sequences in each batch. Or implement a custom sampler that would forbid generating the same data in sequences of the same batch.
> How does structured generation work ?
I used the
Let me know what you think ! And feel free to duplicate/modify it to try other models/prompts or sampling methods :)
features + how it works:
βοΈ Generate the dataset content you want just by entering a file name
π‘ Optionally specify the column names you need
π¨ The dataset is streamed and generated on-the-fly in JSON Lines format
β Generation is constrained to always output valid JSON
How does this work ?
1/ Enter a file name
2/ The model generates column names for such a file. Using structured generation, it can generate 2 to 5 column names using lower case characters and underscores. I use a prompt that asks to generate column names for a realistic dataset and low temperature.
3/ The columns are used to update the Finite State Machine for the dataset content structured generation, so that it is used to generate JSON objects using those columns
4/ The model generates JSON objects using structured generation again, using the updated Finite State Machine. I use a prompt that asks for realistic data and a temperature of 1.
> Why update a Finite State Machine instead of re-creating one ?
Creating one can take up to 30sec, while updating one takes 0.1s (though it requires to manipulate a graph which is not easy to implement)
> Batched generation is faster, why not use it ?
Generate in batches is faster but tends to generate duplicates for this demo.
Further work can be to provide different prompts (one per sequence in the batch) to end up with a different distribution of sequences in each batch. Or implement a custom sampler that would forbid generating the same data in sequences of the same batch.
> How does structured generation work ?
I used the
outlines library with transformers to to define a JSON schema that the generation has to follow. It uses a Finite State Machine with token_id as transitions.Let me know what you think ! And feel free to duplicate/modify it to try other models/prompts or sampling methods :)
anton-lΒ
authored
a
paper
10 months ago
albertvillanovaΒ
updated
a
dataset
11 months ago