ronantakizawa (Ronan Takizawa)

reacted to their post with 👍 about 14 hours ago

Post

400

Introducing JFLEG-JA, a new Japanese language error correction benchmark with 1,335 sentences, each paired with 4 high-quality human corrections 🎉

Inspired by the English JFLEG dataset, this dataset covers diverse error types, including particle mistakes, kanji mix-ups, incorrect contextual verb, adjective, and literary technique usage.

You can use this for evaluating LLMs, few-shot learning, error analysis, or fine-tuning correction systems.

ronantakizawa/jfleg-japanese

#japanese #evals #benchmark

posted an update 1 day ago

Post

400

Introducing JFLEG-JA, a new Japanese language error correction benchmark with 1,335 sentences, each paired with 4 high-quality human corrections 🎉

Inspired by the English JFLEG dataset, this dataset covers diverse error types, including particle mistakes, kanji mix-ups, incorrect contextual verb, adjective, and literary technique usage.

You can use this for evaluating LLMs, few-shot learning, error analysis, or fine-tuning correction systems.

ronantakizawa/jfleg-japanese

#japanese #evals #benchmark

reacted to their post with 👍 5 days ago

Post

1686

Introducing the Medical-o1-Reasoning-SFT-Japanese dataset 🎉

This dataset is a Japanese dataset consisting questions, reasoning, and answer results for complex medical topics.

#japanese #medical #dataset

ronantakizawa/Medical-o1-Reasoning-SFT-Japanese

posted an update 6 days ago

Post

1686

Introducing the Medical-o1-Reasoning-SFT-Japanese dataset 🎉

This dataset is a Japanese dataset consisting questions, reasoning, and answer results for complex medical topics.

#japanese #medical #dataset

ronantakizawa/Medical-o1-Reasoning-SFT-Japanese

reacted to their post with 👍 10 days ago

Post

1467

Introducing the Finance-Instruct-500k-Japanese dataset 🎉

This is a Japanese-translated version of the @Josephgflowers Finance-Instruct-500k dataset, which includes complex questions and answers related to finance and Economics.

#datasets #finance #finance-instruct #japanese

ronantakizawa/Finance-Instruct-500k-Japanese

posted an update 11 days ago

Post

1467

Introducing the Finance-Instruct-500k-Japanese dataset 🎉

This is a Japanese-translated version of the @Josephgflowers Finance-Instruct-500k dataset, which includes complex questions and answers related to finance and Economics.

#datasets #finance #finance-instruct #japanese

ronantakizawa/Finance-Instruct-500k-Japanese

reacted to their post with 🔥 15 days ago

Post

1551

Excited to announce 4 AWQ quantized models from #AllenAI! 🎉

Molmo-7B-D AWQ (14GB→5GB): Efficient VLM performing between GPT-4V and GPT-4o on academic benchmarks, with just 6.1% perplexity degradation.

MolmoAct-7B-D AWQ (14GB→6GB): Specialized robotic manipulation model reduced by ~57%.

Molmo-72B AWQ (145GB→38GB): VLM with Qwen2-72B decoder that performs competitively with GPT-4, achieving only 10.5% perplexity degradation while saving 107GB of memory.

OLMo-2-32B-Instruct AWQ (64GB→17GB): LLM post-trained on Tülu 3 with 3% perplexity degradation while saving ~50GB.

All VLMs only had their text models quantized.

ronantakizawa/molmo-7b-d-awq
ronantakizawa/molmoact-7b-d-awq
ronantakizawa/molmo-72b-awq
ronantakizawa/olmo2-32b-instruct-awq

posted an update 16 days ago

Post

1551

Excited to announce 4 AWQ quantized models from #AllenAI! 🎉

Molmo-7B-D AWQ (14GB→5GB): Efficient VLM performing between GPT-4V and GPT-4o on academic benchmarks, with just 6.1% perplexity degradation.

MolmoAct-7B-D AWQ (14GB→6GB): Specialized robotic manipulation model reduced by ~57%.

Molmo-72B AWQ (145GB→38GB): VLM with Qwen2-72B decoder that performs competitively with GPT-4, achieving only 10.5% perplexity degradation while saving 107GB of memory.

OLMo-2-32B-Instruct AWQ (64GB→17GB): LLM post-trained on Tülu 3 with 3% perplexity degradation while saving ~50GB.

All VLMs only had their text models quantized.

ronantakizawa/molmo-7b-d-awq
ronantakizawa/molmoact-7b-d-awq
ronantakizawa/molmo-72b-awq
ronantakizawa/olmo2-32b-instruct-awq

reacted to their post with 🚀👍 19 days ago

Post

3808

Introducing AWQ and GPTQ quantized versions of SmolVLM from Hugging Face!

These models only had their text models quantized, and had a 50% model size reduction (4GB~2GB) while keeping model degradation under 1% on the DocVQA benchmark.

#huggingface #smolvlm #smollm

ronantakizawa/SmolVLM-Instruct-awq

ronantakizawa/SmolVLM-Instruct-gptq

posted an update 19 days ago

Post

3808

Introducing AWQ and GPTQ quantized versions of SmolVLM from Hugging Face!

These models only had their text models quantized, and had a 50% model size reduction (4GB~2GB) while keeping model degradation under 1% on the DocVQA benchmark.

#huggingface #smolvlm #smollm

ronantakizawa/SmolVLM-Instruct-awq

ronantakizawa/SmolVLM-Instruct-gptq

reacted to their post with 👀 23 days ago

Post

3428

Released an AWQ quantized version of BosonAI’s Higgs-Llama-3-70B model! 🎉
The Higgs-Llama-3-70B is an LLM specialized in role-playing, useful for game characters.

Using an NVIDIA B200 GPU, I was able to compress the huge 140GB model into 37GB while keeping minimal perplexity 👍

ronantakizawa/higgs-llama-3-70b-awq

posted an update 23 days ago

Post

3428

Released an AWQ quantized version of BosonAI’s Higgs-Llama-3-70B model! 🎉
The Higgs-Llama-3-70B is an LLM specialized in role-playing, useful for game characters.

Using an NVIDIA B200 GPU, I was able to compress the huge 140GB model into 37GB while keeping minimal perplexity 👍

ronantakizawa/higgs-llama-3-70b-awq

reacted to their post with 🔥👍 24 days ago

Post

1825

Released a 4-bit quantized version of Microsoft's Phi-4-reasoning model using Activation-aware Weight Quantization (AWQ)!

#phi-4 #quantization

ronantakizawa/phi-4-reasoning-awq

2 replies

·

reacted to their post with 👍 27 days ago

Post

966

Introducing the japanese-text-difficulty dataset! 🎉

This dataset gathered texts from Aozora Bunko and marked them with jReadability scores, plus detailed metrics on kanji density, vocabulary, grammar, and sentence structure.

This is an excellent dataset if you want to train your LLM to understand the complexities of the Japanese language.

ronantakizawa/japanese-text-difficulty

#dataset #japanese #textdifficulty

posted an update 28 days ago

Post

966

Introducing the japanese-text-difficulty dataset! 🎉

This dataset gathered texts from Aozora Bunko and marked them with jReadability scores, plus detailed metrics on kanji density, vocabulary, grammar, and sentence structure.

This is an excellent dataset if you want to train your LLM to understand the complexities of the Japanese language.

ronantakizawa/japanese-text-difficulty

#dataset #japanese #textdifficulty

posted an update about 1 month ago

Post

1825

Released a 4-bit quantized version of Microsoft's Phi-4-reasoning model using Activation-aware Weight Quantization (AWQ)!

#phi-4 #quantization

ronantakizawa/phi-4-reasoning-awq

2 replies

·

Ronan Takizawa PRO

AI & ML interests

Recent Activity

Organizations

Ronan Takizawa PRO

AI & ML interests

Recent Activity

Organizations

ronantakizawa's activity