Taylor Stacey
@taylorstacey.ca
Probably doing something GKE or LLM related these days. Can rally at Canadian trivia nights.
AI Architect @ SADA
Vancouver, Canada
AI Architect @ SADA
Vancouver, Canada
Reposted by Taylor Stacey
Incredibly moving Justin Trudeau remarks:
"We have fought and died alongside you....During your darkest hours...we were always there. Standing with you, grieving with you, the American people....Canadians are a little perplexed as to why our closest friends and neighbors are choosing to target us."
"We have fought and died alongside you....During your darkest hours...we were always there. Standing with you, grieving with you, the American people....Canadians are a little perplexed as to why our closest friends and neighbors are choosing to target us."
February 2, 2025 at 3:39 AM
Incredibly moving Justin Trudeau remarks:
"We have fought and died alongside you....During your darkest hours...we were always there. Standing with you, grieving with you, the American people....Canadians are a little perplexed as to why our closest friends and neighbors are choosing to target us."
"We have fought and died alongside you....During your darkest hours...we were always there. Standing with you, grieving with you, the American people....Canadians are a little perplexed as to why our closest friends and neighbors are choosing to target us."
Not super well documented but the google-cloud-platform SDK will let you pass in your own headers with request_metadata. This means you can force future generate_content() calls to only use provisioned throughput.
`vertexai.init(request_metadata=[("x-vertex-ai-llm-request-type","dedicated")])`
`vertexai.init(request_metadata=[("x-vertex-ai-llm-request-type","dedicated")])`
January 7, 2025 at 10:39 PM
Not super well documented but the google-cloud-platform SDK will let you pass in your own headers with request_metadata. This means you can force future generate_content() calls to only use provisioned throughput.
`vertexai.init(request_metadata=[("x-vertex-ai-llm-request-type","dedicated")])`
`vertexai.init(request_metadata=[("x-vertex-ai-llm-request-type","dedicated")])`
The Self Care Beach Boys Playlist
December 30, 2024 at 8:26 PM
The Self Care Beach Boys Playlist
It’s sure not the antenna it came with but very pleased I repaired it myself
December 5, 2024 at 3:57 AM
It’s sure not the antenna it came with but very pleased I repaired it myself
Do I know anyone who's been enriching long videos with VLMs, like Gemini and soon Nova? Would love to pick your brain.
December 4, 2024 at 8:32 PM
Do I know anyone who's been enriching long videos with VLMs, like Gemini and soon Nova? Would love to pick your brain.
Hotels in Vancouver are notifying customers that they’ve overbooked and are cancelling reservations. I would bet so much money they thought “Let’s see how much we can get?” and then booted the folks whom booked early.
December 4, 2024 at 8:31 PM
Hotels in Vancouver are notifying customers that they’ve overbooked and are cancelling reservations. I would bet so much money they thought “Let’s see how much we can get?” and then booted the folks whom booked early.
Deprecating AutoML for text makes sense. AutoML for video still lives, which doesn't surprise me given Gemini's 1 fps downsampling for video content. Curious if and when the downsampling will be removed.
November 28, 2024 at 6:30 PM
Deprecating AutoML for text makes sense. AutoML for video still lives, which doesn't surprise me given Gemini's 1 fps downsampling for video content. Curious if and when the downsampling will be removed.
The 92% discount on Mass Effect Legendary is nuts. Amazing value.
November 28, 2024 at 4:15 AM
The 92% discount on Mass Effect Legendary is nuts. Amazing value.
Another beautiful sunset off kits beach
November 28, 2024 at 12:19 AM
Another beautiful sunset off kits beach
November 27, 2024 at 8:59 PM
We're like 99% there on Agent Builder. Until then.
November 27, 2024 at 8:16 PM
We're like 99% there on Agent Builder. Until then.
Not that I need to know I'm valid but I'd like the reassurance.
November 27, 2024 at 7:04 PM
Not that I need to know I'm valid but I'd like the reassurance.
Still one of the more interesting finds with LLMs. They sure like clarity. When given two tables with the same data, the LLM's chain of thought shows the model decided to choose the table that actually provided details in the table schema.
November 27, 2024 at 6:53 PM
Still one of the more interesting finds with LLMs. They sure like clarity. When given two tables with the same data, the LLM's chain of thought shows the model decided to choose the table that actually provided details in the table schema.
