QVQ 72B Preview - a Hugging Face Space by Qwen
Discover amazing ML apps made by the community
huggingface.co
December 24, 2024 at 5:26 PM
Model weights here: huggingface.co/Qwen/QVQ-72B...
Qwen/QVQ-72B-Preview · Hugging Face
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
huggingface.co
December 24, 2024 at 5:25 PM
Model weights here: huggingface.co/Qwen/QVQ-72B...
3.1 70B vs 3.3 70B:
Code Generation
> HumanEval: 80.5% → 88.4% (+7.9%)
> MBPP EvalPlus: 86.0% → 87.6% (+1.6%)
Steerability
> IFEval: 87.5% → 92.1% (+4.6%)
Reasoning & Math
> GPQA Diamond (CoT): 48.0% → 50.5% (+2.5%)
> MATH (CoT): 68.0% → 77.0% (+9%)
Code Generation
> HumanEval: 80.5% → 88.4% (+7.9%)
> MBPP EvalPlus: 86.0% → 87.6% (+1.6%)
Steerability
> IFEval: 87.5% → 92.1% (+4.6%)
Reasoning & Math
> GPQA Diamond (CoT): 48.0% → 50.5% (+2.5%)
> MATH (CoT): 68.0% → 77.0% (+9%)
December 6, 2024 at 6:20 PM
3.1 70B vs 3.3 70B:
Code Generation
> HumanEval: 80.5% → 88.4% (+7.9%)
> MBPP EvalPlus: 86.0% → 87.6% (+1.6%)
Steerability
> IFEval: 87.5% → 92.1% (+4.6%)
Reasoning & Math
> GPQA Diamond (CoT): 48.0% → 50.5% (+2.5%)
> MATH (CoT): 68.0% → 77.0% (+9%)
Code Generation
> HumanEval: 80.5% → 88.4% (+7.9%)
> MBPP EvalPlus: 86.0% → 87.6% (+1.6%)
Steerability
> IFEval: 87.5% → 92.1% (+4.6%)
Reasoning & Math
> GPQA Diamond (CoT): 48.0% → 50.5% (+2.5%)
> MATH (CoT): 68.0% → 77.0% (+9%)
Llama 3.3 70B vs 405B:
> GPQA Diamond (CoT): 50.5% vs 49.0%
> Math (CoT): 77.0% vs 73.8%
> Steerability (IFEval): 92.1% vs 88.6%
huggingface.co/meta-llama/L...
> GPQA Diamond (CoT): 50.5% vs 49.0%
> Math (CoT): 77.0% vs 73.8%
> Steerability (IFEval): 92.1% vs 88.6%
huggingface.co/meta-llama/L...
meta-llama/Llama-3.3-70B-Instruct · Hugging Face
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
huggingface.co
December 6, 2024 at 6:20 PM
Llama 3.3 70B vs 405B:
> GPQA Diamond (CoT): 50.5% vs 49.0%
> Math (CoT): 77.0% vs 73.8%
> Steerability (IFEval): 92.1% vs 88.6%
huggingface.co/meta-llama/L...
> GPQA Diamond (CoT): 50.5% vs 49.0%
> Math (CoT): 77.0% vs 73.8%
> Steerability (IFEval): 92.1% vs 88.6%
huggingface.co/meta-llama/L...
Indic Parler-TTS - a Hugging Face Space by ai4bharat
A demo of Indic Parler-TTS
huggingface.co
December 3, 2024 at 9:33 PM
try it out today on hf.co/datasets - just click on `SQL Console` followed by `AI Query` 💯
Hugging Face – The AI community building the future.
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
hf.co
December 2, 2024 at 2:42 PM
try it out today on hf.co/datasets - just click on `SQL Console` followed by `AI Query` 💯
To showcase how much you can do with just a 1.7B LLM, you pass free text, define a schema of parsing the text into a GitHub issue (title, description, categories, tags, etc) - Let MLC & XGrammar do the rest!
That's it, the code is super readable, try it out today! 🤗
huggingface.co/spaces/reach...
That's it, the code is super readable, try it out today! 🤗
huggingface.co/spaces/reach...
Github Issue Generator - a Hugging Face Space by reach-vb
Discover amazing ML apps made by the community
huggingface.co
November 28, 2024 at 10:25 PM
To showcase how much you can do with just a 1.7B LLM, you pass free text, define a schema of parsing the text into a GitHub issue (title, description, categories, tags, etc) - Let MLC & XGrammar do the rest!
That's it, the code is super readable, try it out today! 🤗
huggingface.co/spaces/reach...
That's it, the code is super readable, try it out today! 🤗
huggingface.co/spaces/reach...
> uses 90% sliding window and 10% global attention for efficiency
> 2-stage pre-training and 3-phase post-training, including a trapezoid learning rate schedule
try it out on hugging face today! 🤗
huggingface.co/collections/...
> 2-stage pre-training and 3-phase post-training, including a trapezoid learning rate schedule
try it out on hugging face today! 🤗
huggingface.co/collections/...
Hymba - a nvidia Collection
A series of Hybrid Small Language Models.
huggingface.co
November 26, 2024 at 7:34 PM
> uses 90% sliding window and 10% global attention for efficiency
> 2-stage pre-training and 3-phase post-training, including a trapezoid learning rate schedule
try it out on hugging face today! 🤗
huggingface.co/collections/...
> 2-stage pre-training and 3-phase post-training, including a trapezoid learning rate schedule
try it out on hugging face today! 🤗
huggingface.co/collections/...
Model weights on the hub, you can even run this on a Raspberry Pi! Go run, inference now! 🐐
huggingface.co/OuteAI/OuteT...
huggingface.co/OuteAI/OuteT...
OuteAI/OuteTTS-0.2-500M · Hugging Face
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
huggingface.co
November 25, 2024 at 9:33 PM
Model weights on the hub, you can even run this on a Raspberry Pi! Go run, inference now! 🐐
huggingface.co/OuteAI/OuteT...
huggingface.co/OuteAI/OuteT...