dahara1
@dahara1.bsky.social
I made machine translation with LLMs.
I made PC Chrome translation plugin for bluesky.
I made smart feed for bluesky.
I mada content import agent.
Let's improve these qualities!
I made PC Chrome translation plugin for bluesky.
I made smart feed for bluesky.
I mada content import agent.
Let's improve these qualities!
Pinned
dahara1
@dahara1.bsky.social
· Jul 26
VoiceCore - AI音声生成システム
webbigdata.jp
We have finally completed a TTS model that can generate emotional Japanese speech from text.
Those who can speak Japanese might be interested.
webbigdata.jp/voice-ai-age...
Those who can speak Japanese might be interested.
webbigdata.jp/voice-ai-age...
@dorialexander.bsky.social
You may not remember, but just in case.
An organization that includes Studio Ghibli has submitted a request to OpenAI's Sora2 not to use it as a learning target without permission.
coda-cj.jp/news/2577/
You may not remember, but just in case.
An organization that includes Studio Ghibli has submitted a request to OpenAI's Sora2 not to use it as a learning target without permission.
coda-cj.jp/news/2577/
OpenAI社に「Sora 2」の運用に関する要望書を提出 | 一般社団法人コンテンツ海外流通促進機構（CODA）
CODAは会員社からの要請に基づき、2025年10月27日、OpenAI社に対し、同社が2025年9月30日
coda-cj.jp
October 30, 2025 at 10:26 AM
@dorialexander.bsky.social
You may not remember, but just in case.
An organization that includes Studio Ghibli has submitted a request to OpenAI's Sora2 not to use it as a learning target without permission.
coda-cj.jp/news/2577/
You may not remember, but just in case.
An organization that includes Studio Ghibli has submitted a request to OpenAI's Sora2 not to use it as a learning target without permission.
coda-cj.jp/news/2577/
The Japanese version of nanochat's middle training will be available soon.
October 22, 2025 at 1:27 PM
The Japanese version of nanochat's middle training will be available soon.
Unless you're a famous person, your website or blog won't be respected or cited.
They're treated as "data picked up from the internet," and crawlers will just occasionally visit them.
In the age of AI, will we be forced to participate in the attention economy, such as SNS and video streaming?
They're treated as "data picked up from the internet," and crawlers will just occasionally visit them.
In the age of AI, will we be forced to participate in the attention economy, such as SNS and video streaming?
October 10, 2025 at 5:35 AM
Unless you're a famous person, your website or blog won't be respected or cited.
They're treated as "data picked up from the internet," and crawlers will just occasionally visit them.
In the age of AI, will we be forced to participate in the attention economy, such as SNS and video streaming?
They're treated as "data picked up from the internet," and crawlers will just occasionally visit them.
In the age of AI, will we be forced to participate in the attention economy, such as SNS and video streaming?
I feel like Claude is sometimes better than Gemini at using OSS tools uploaded to GitHub.
Gemini has a lot of hallucinations when it comes to GitHub.
Gemini has a lot of hallucinations when it comes to GitHub.
October 9, 2025 at 5:08 PM
I feel like Claude is sometimes better than Gemini at using OSS tools uploaded to GitHub.
Gemini has a lot of hallucinations when it comes to GitHub.
Gemini has a lot of hallucinations when it comes to GitHub.
I chatted with Claude about forgetting some of the settings I'd made in the past, and he automatically searched my chat history.
This is more convenient than searching through chat history.
This is more convenient than searching through chat history.
September 24, 2025 at 3:12 AM
I chatted with Claude about forgetting some of the settings I'd made in the past, and he automatically searched my chat history.
This is more convenient than searching through chat history.
This is more convenient than searching through chat history.
Gemini Pro writes the modified code as follows:
# ★ Add from here
# ★ Add up to here
I found it very useful.
However, Be carefuel!
I found that it can be extra work because Gemini comments are sometimes missed.
# ★ Add from here
# ★ Add up to here
I found it very useful.
However, Be carefuel!
I found that it can be extra work because Gemini comments are sometimes missed.
September 14, 2025 at 11:10 AM
Gemini Pro writes the modified code as follows:
# ★ Add from here
# ★ Add up to here
I found it very useful.
However, Be carefuel!
I found that it can be extra work because Gemini comments are sometimes missed.
# ★ Add from here
# ★ Add up to here
I found it very useful.
However, Be carefuel!
I found that it can be extra work because Gemini comments are sometimes missed.
LLM Development
Even publicly available foundational models always have some kind of bug.
Even well-known frameworks always have some kind of bug.
Even proven libraries can suddenly stop working.
Datasets always contain some kind of abnormal data.
Conventional wisdom changes at a rapid pace
Even publicly available foundational models always have some kind of bug.
Even well-known frameworks always have some kind of bug.
Even proven libraries can suddenly stop working.
Datasets always contain some kind of abnormal data.
Conventional wisdom changes at a rapid pace
September 13, 2025 at 5:28 AM
LLM Development
Even publicly available foundational models always have some kind of bug.
Even well-known frameworks always have some kind of bug.
Even proven libraries can suddenly stop working.
Datasets always contain some kind of abnormal data.
Conventional wisdom changes at a rapid pace
Even publicly available foundational models always have some kind of bug.
Even well-known frameworks always have some kind of bug.
Even proven libraries can suddenly stop working.
Datasets always contain some kind of abnormal data.
Conventional wisdom changes at a rapid pace
What you quickly discover when you challenge an AI agent to search the web.
Google search is truly amazing.
Google search is truly amazing.
September 13, 2025 at 3:28 AM
What you quickly discover when you challenge an AI agent to search the web.
Google search is truly amazing.
Google search is truly amazing.
More and more people are using pre-training or continuous pre-training for SLM, but it's really difficult.
Running it on a consumer-grade GPU isn't hard, but it takes over 1,000 hours. Even then, the data volume is insufficient, resulting in a dumb model.
Running it on a consumer-grade GPU isn't hard, but it takes over 1,000 hours. Even then, the data volume is insufficient, resulting in a dumb model.
September 7, 2025 at 9:56 AM
More and more people are using pre-training or continuous pre-training for SLM, but it's really difficult.
Running it on a consumer-grade GPU isn't hard, but it takes over 1,000 hours. Even then, the data volume is insufficient, resulting in a dumb model.
Running it on a consumer-grade GPU isn't hard, but it takes over 1,000 hours. Even then, the data volume is insufficient, resulting in a dumb model.
EmbeddingGemma was faster than BERT when run on a CPU! BERT was faster on a GPU.
This really seems like it would be good for edge computing.
This really seems like it would be good for edge computing.
September 5, 2025 at 8:29 AM
EmbeddingGemma was faster than BERT when run on a CPU! BERT was faster on a GPU.
This really seems like it would be good for edge computing.
This really seems like it would be good for edge computing.
The AI revolution has begun.
September 4, 2025 at 4:47 PM
The AI revolution has begun.
(1)Easy plan
Instruct model + finetune
(2)Solid plan
Base model + Continuous pre training + post training + finetune
(3)Advanced plan
Base model + Continuous pre training + (Instruct model - Base model) + post training + finetune
Instruct model + finetune
(2)Solid plan
Base model + Continuous pre training + post training + finetune
(3)Advanced plan
Base model + Continuous pre training + (Instruct model - Base model) + post training + finetune
September 1, 2025 at 12:31 PM
(1)Easy plan
Instruct model + finetune
(2)Solid plan
Base model + Continuous pre training + post training + finetune
(3)Advanced plan
Base model + Continuous pre training + (Instruct model - Base model) + post training + finetune
Instruct model + finetune
(2)Solid plan
Base model + Continuous pre training + post training + finetune
(3)Advanced plan
Base model + Continuous pre training + (Instruct model - Base model) + post training + finetune
If the model places too much emphasis on safety, it will likely start to ignore user instructions.
But who will check the output of an AI agent that doesn't follow instructions?
Will humans be responsible for checking it?
But who will check the output of an AI agent that doesn't follow instructions?
Will humans be responsible for checking it?
September 1, 2025 at 4:20 AM
If the model places too much emphasis on safety, it will likely start to ignore user instructions.
But who will check the output of an AI agent that doesn't follow instructions?
Will humans be responsible for checking it?
But who will check the output of an AI agent that doesn't follow instructions?
Will humans be responsible for checking it?
gpt-oss is fast, but it seems to have poor ability to follow detailed instructions.
Is it better to use it assuming the use of tools?
Is it better to use it assuming the use of tools?
August 31, 2025 at 2:48 AM
gpt-oss is fast, but it seems to have poor ability to follow detailed instructions.
Is it better to use it assuming the use of tools?
Is it better to use it assuming the use of tools?
Surprisingly, Gemini didn't get the joke just by looking at it, but Claude did.
The title is "Italians are furious."
The title is "Italians are furious."
August 26, 2025 at 3:01 PM
Surprisingly, Gemini didn't get the joke just by looking at it, but Claude did.
The title is "Italians are furious."
The title is "Italians are furious."
There are rumors that llama 4.1 and 4.2 are SLM.
August 15, 2025 at 8:58 AM
There are rumors that llama 4.1 and 4.2 are SLM.
An increasing number of services and products are setting up "AI support chatbots" without publishing product manuals or usage instructions on their websites.
Without reliable documentation, the AI's responses will be hallucinatory and completely useless.
Without reliable documentation, the AI's responses will be hallucinatory and completely useless.
August 12, 2025 at 4:49 AM
An increasing number of services and products are setting up "AI support chatbots" without publishing product manuals or usage instructions on their websites.
Without reliable documentation, the AI's responses will be hallucinatory and completely useless.
Without reliable documentation, the AI's responses will be hallucinatory and completely useless.
As AI has made writing easier, contests and other events have begun requiring the submission of explanatory videos.
Overall, it seems like there's more work for humans to do than ever before.
Overall, it seems like there's more work for humans to do than ever before.
August 7, 2025 at 2:53 AM
As AI has made writing easier, contests and other events have begun requiring the submission of explanatory videos.
Overall, it seems like there's more work for humans to do than ever before.
Overall, it seems like there's more work for humans to do than ever before.
Opus can no longer do the tasks that it was able to do a few weeks ago. I'm very sad.
August 2, 2025 at 10:40 AM
Opus can no longer do the tasks that it was able to do a few weeks ago. I'm very sad.
AI winter is coming.
I introduced VoiceCore, LLM-based Japanese TTS I created, to the Japanese-learning community, but reaction was negative.
People are getting tired of the innovative AI-powered learning materials introduced by influencers on TikTok, and anything with the word "AI" is boring.
I introduced VoiceCore, LLM-based Japanese TTS I created, to the Japanese-learning community, but reaction was negative.
People are getting tired of the innovative AI-powered learning materials introduced by influencers on TikTok, and anything with the word "AI" is boring.
July 31, 2025 at 6:53 AM
AI winter is coming.
I introduced VoiceCore, LLM-based Japanese TTS I created, to the Japanese-learning community, but reaction was negative.
People are getting tired of the innovative AI-powered learning materials introduced by influencers on TikTok, and anything with the word "AI" is boring.
I introduced VoiceCore, LLM-based Japanese TTS I created, to the Japanese-learning community, but reaction was negative.
People are getting tired of the innovative AI-powered learning materials introduced by influencers on TikTok, and anything with the word "AI" is boring.
We have finally completed a TTS model that can generate emotional Japanese speech from text.
Those who can speak Japanese might be interested.
webbigdata.jp/voice-ai-age...
Those who can speak Japanese might be interested.
webbigdata.jp/voice-ai-age...
VoiceCore - AI音声生成システム
webbigdata.jp
July 26, 2025 at 7:06 AM
We have finally completed a TTS model that can generate emotional Japanese speech from text.
Those who can speak Japanese might be interested.
webbigdata.jp/voice-ai-age...
Those who can speak Japanese might be interested.
webbigdata.jp/voice-ai-age...
Opus omits just two lines of main, saying "the rest of the code is the same"
↓
I spent two hours debugging with Gemini to find out why the app suddenly stopped working
↓
Rage with nowhere to go
↓
I spent two hours debugging with Gemini to find out why the app suddenly stopped working
↓
Rage with nowhere to go
July 25, 2025 at 2:02 PM
Opus omits just two lines of main, saying "the rest of the code is the same"
↓
I spent two hours debugging with Gemini to find out why the app suddenly stopped working
↓
Rage with nowhere to go
↓
I spent two hours debugging with Gemini to find out why the app suddenly stopped working
↓
Rage with nowhere to go
It's hard to find a single prompt that will always give you the perfect answer.
You might want to consider splitting the answer and the verification into two prompts.
You might want to consider splitting the answer and the verification into two prompts.
July 25, 2025 at 5:17 AM
It's hard to find a single prompt that will always give you the perfect answer.
You might want to consider splitting the answer and the verification into two prompts.
You might want to consider splitting the answer and the verification into two prompts.
