Toronto, Ontario
- Tweets 2,236
- Following 824
- Followers 2,419
- Likes 24,040
"build the world we want to live in" is a simple mission. For me and for Ambition, that means celebrating the pursuit of depth and understanding, making people believe that they can be ambitious in ways they thought they couldn't be before
Ambition's mission is simple: build the world we want to live in. That entails building a culture that celebrates the pursuit of depth and understanding
2
9
52
11,912
I love Toronto but sometimes people just need to say no to things. 40 people signing up for something and 10 showing up is just embarrassing for us, collectively. Please respect yourself and your time and just say no.
26
32
597
61,322
Prime intellect probably has my favourite branding out of all the AI companies out there
6
12
481
49,423
Excited to start at @runwayml this week! Only three days in, and it’s already clear how fast and focused the team is. Feels good to work on something that brings together AI and filmmaking—two things I really care about. lfg
17
1
180
11,849
Announcing The Toronto School Of Foundation Modelling, a Toronto exclusive, in-person only school for learning to build Foundation Models. Coming to New Stadium and Youthful Vengeance in late August 2025.
25
14
178
28,003
Birthday 2019: Canadian visitor visa to attend an ML fellowship in Toronto gets rejected
Birthday 2024: Canadian citizenship application gets approved
What a journey
24
3
154
8,934
Idk why Torontonians keep hating on Drake. He’s the one big artist who actually represents the city, goes to the games to support the teams and has built an identity around living in Toronto. Makes no sense.
16
4
130
10,254
Thrilled to welcome @modal as the compute sponsor of Toronto School of Foundation Modelling! We’ll be using Modal Notebooks: GPU-backed Python in the browser that starts in seconds, supports real-time collaboration so this cohort can start experimenting right away. More next week
6
7
107
21,345
If you're passionate about crafting an unbeatable experience for training LLMs at scale and want to build the best LLM finetuning experience in the industry, consider applying! I'm hiring for my team in Toronto that focuses on precisely that (DMs are open)
jobs.lever.co/cohere/f36be9d…
6
15
86
17,638
Thank you to everyone who showed up to the first lecture of TSFM. So excited to be on this journey with all of you! Will refine the format over the series based on your feedback and make it good :)))
See you next Thursday
6
7
68
3,531
It’s my last week at @runwayml. Grateful for an exceptionally kind, driven and talented team that taught me something new almost everyday. I'm taking a short break to focus on my health (and to catch up on my bookmarks folder). You'll see me doing stuff soon enough.
6
68
3,574
22 lectures, 70 hours, 12 speakers, 60 students, 8 sponsors and 2 venues.
The Toronto School of Foundation Modelling is coming to an end. Celebrate it with me
luma.com/bqjd2j2n
3
7
74
11,827
🥹🥹
I will do everything I can to get people to discover the top end of what they can learn and do.
1
2
63
2,466
For better or for worse, I’ve always viewed money as a tool to do things. I didn’t grow up rich but I saw my parents do things and be generous where it matters. So I do the same. I’m not rich but I run TSFM which is a fair bit of expenses out of my own pocket, I produced 5 short films and a feature length documentary and I helped start Youthful Vengeance. It just bothers me when I see tech people who make so much save up a lot, for nothing real. Spend it. Money is a tool. Make things happen using it.
2
2
61
2,508
Anyone that is excellent at RL for LLMs that is willing to give a guest lecture at Toronto School of Foundation Modelling? I will cover flights, accommodation and introduce you to the best food in the world
7
8
61
8,039
The TSFM cohort showed up on a Friday night to learn about distributed training for 3 hours. And then reconvened on Saturday morning to learn about AlphaFold and drug discovery. Both talks are deeply technical yet it was a full house. Bullish on Toronto and its people
8
5
61
3,123
Spent the last week in the Canadian Rockies with my best friends. This country is so beautiful
1
2
58
2,169
Last two weeks at the Toronto School of Foundation Modelling:
- Building a multimodal data pipeline by Cindy and Ethan from Runway.
- Scaling Drug Discovery Search Space by Joanna from Isomorphic Labs.
3
59
3,993
The world’s best open weight model is here
Today, we’re introducing Command R+: a state-of-the-art RAG-optimized LLM designed to tackle enterprise-grade workloads and speak the languages of global business.
Our R-series model family is now available on Microsoft Azure, and coming soon to additional cloud providers.
4
51
3,340
I showed up to Stadium today at noon, feeling exhausted. Over the course of five hours, I had such great conversations with the people there that it energized me for TSFM03. Truly a special day at Stadium, grateful that the space exists.
6
6
56
3,583
Season 1 of Toronto School of Foundation Modelling kicks off this Thursday at New Stadium!!!
60 people will be attending weekly sessions for 3 months, learning to build Foundation Models from scratch. Around 10 guest speakers (more to come) will be flying to Toronto to talk about what they do best.
I'm grateful for the support of Cohere, New, Vengeance and all the donors. I will make this a series worth your time. And apologies to the people I didn't get back to - spots are all taken but will reach out if that changes.
Continuing to press forward with the range and depth of learning opportunities.
This week we have several workshops, meet-ups, a deep-dive seminar, the beginning of a new lecture series, as well as an exhibit happening at New Stadium.
Links below.
4
13
55
7,240
Working on a side project with Claude called Granter that takes in info about your org, scans through available grants and ranks them based on compatibility with your org's mission, generates drafts for those you're interested in. Late to the party but crazy this took ~3hrs
7
3
56
3,780
Trying to raise one last batch of funds for season 1 of Toronto School of Foundation Modelling! We are nearing the end of the first season with many in the cohort already training their own models, discovering scaling laws, building tokenizers etc. I’m planning a few surprises for the end which could use additional funding. If you’re interested in supporting, please reach out or go to tsfm dot ca / donate directly.
3
13
51
3,946
Come train neural nets with me
🐾PAWLEAGUE APPRECIATION POST
Did you know that all of our offices are dog-friendly? When we're not training large language models, we're training four-legged co:mpanions (or maybe the other way around). We've even had true friendship blossom between @chengstaa & Nicole's doggos.
3
3
50
15,833
New radicalized me. Incredible things can and will happen when you run a space that hosts curious people. My definition of success has broadened because I saw people pushing their own limits and define what excellence is to them
1
5
51
3,432
Shoutout @ftnabeelah for connecting me with so many wonderful people! Very grateful for her presence in Toronto
8
46
4,040
Every Thursday, I leave home on a mission to understand and teach how to build foundation models in the presence of some of the most driven people I have ever met. Behind the scenes are 20 chairs, one macbook and a few friends who help me realize this mission. More BTS of TSFM soon
47
2,117
More visualizations I built for TSFM (lectures 3 and 4)
1. FLOPs allocation for vanilla transformer
2. Visual explanation for RoPE
3. Suboptimal ops for attention on H100 - to motivate GPU efficiency imp
4. Gradient norms for last layer in Pre-Norm vs Post-Norm
1
3
44
1,955
Been incredible watching @acyr_l and his team work to push our base models forward. Onwards and upwards, so excited to build and ship more this year
Today, we’re introducing Command R+: a state-of-the-art RAG-optimized LLM designed to tackle enterprise-grade workloads and speak the languages of global business.
Our R-series model family is now available on Microsoft Azure, and coming soon to additional cloud providers.
4
1
43
2,373
Was discussing RoPE at the TSFM lecture yesterday and ppl had similar experiences. It is a crazy idea. One of the audience members shared this in the slack group:
nature.com/articles/s41586-0…
RoPE has interesting parallels to how position is encoded in biological NNs.
3
5
43
1,465
There is a guy that will be commuting from Montreal to attend TSFM starting next week. It makes no sense but he is willing to do it to meet the cohort and learn in-person every week.
I have so much respect for everyone in TSFM. Showing up every week, balancing full-time jobs or undergrad/master’s programs, sitting through lectures, and then grinding on assignments. I don't think I could do it.
Four weeks in, and it’s obvious: you have to be wired a certain way to commit to something this demanding. The level of seriousness in this group is inspiring. I know I’ll come away having learned as much from them as from the program itself.
3
1
41
2,566
We play to win
We’re excited to announce our Series D financing to accelerate growth, expand our team, & develop our next class of frontier, enterprise-grade, data privacy-focused AI technology.
We’re bringing highly scalable and secure AI solutions to global enterprises wherever their data is and in whatever language they speak.
1
42
4,696
Undergrad and master’s curriculums should have a course that highlights the importance of good engineering and infra in the context of doing research. I feel like very few people that are just out of school (incl me) appreciate what good eng+infra can enable for research at scale
1
1
38
I will not stop until TSFM becomes the best program in the world for learning how to build frontier models.
Lot of work to be done, lot of stuff for me to learn
More confirmed guest lectures:
- Humans of AI: Annotations for LLMs by @chengstaa from Cohere
- Scaling search space by [REDACTED] from Isomorphic Labs
Only at Toronto School of Foundation Modelling.
2
1
40
5,629
RRR (Rejuvenated Refreshed R)
We’re releasing improved versions of the Command R series, our enterprise-grade AI models optimized for business use cases.
You can access them on our API, @awscloud Sagemaker, and additional platforms soon.
cohere.com/blog/command-seri…
1
1
40
2,990
Season 1 of Toronto School of Foundation Modelling is wrapping up soon! 🥹The graduation party will be held on November 27 and will be open to public (more details to come soon). Very grateful for @transformerlab and @aliasaria’s support in making the event possible!!!
1
6
40
2,345
Really glad that the plan for how I’m going to spend the next decade is clear
Yesterday: took the citizenship oath
2 years from now: join the Canadian Space Agency
10 years from now: Quit CSA, go back to India and build infrastructure in villages
iykyk
5
2
37
3,213
People are travelling from Windsor and Waterloo to attend TSFM, it's crazy. I have to make this worth their time and I will. If it means I have to pull all-nighters to teach myself things, so be it man
2
1
39
2,790
Excited to announce @boomvideoapp as the graduation sponsor of Toronto School of Foundation Modelling! I use Boom to record most of my walkthroughs and lectures so I’m extra grateful for their support.
2
39
6,623
I unironically thrive in Toronto summers winters. The colder it gets, the more excited I get. But here’s the catch, I’m dumb as fuck
1
1
37
2,566
Good upcoming week for technical lectures in Toronto
September 16: Human Annotations for LLMs at Youthful Vengeance
September 17: RL verifiers at Studio 535
September 18: GPUs and TPUs at New Stadium
1
2
38
1,639
This Thursday at TSFM:
Transformers Part 2 by one of the best engineers I know, Bharat Venkatesh from @cohere . Bharat took various parts of the Cohere tech stack from 0 to 1 (sometimes singlehandedly!!) over the course of his 5.5 years at Cohere. So excited to learn from him!!
4
38
2,118
Hosting a live Q&A / discussion about the Toronto School Of Foundation Modelling at New Stadium this Friday at 1pm. Will be talking about structure, curriculum and vision
1
4
37
3,169
Honoured to be hosting @1vnzh for a fireside chat at the TSFM end-of-season party on November 27. TSFM exists because of the early belief and support from Ivan and Cohere - can’t think of a better way to close out season 1. More reflections soon.
2
38
1,418
RoPE’s behaviour is very perplexing. It’s almost universally used but its effects are poorly understood (much like the rest of LLMs I suppose). I am almost uncomfortable using it. For some reason, 10k is the worst theta to use, 500 is very good for extrapolation but not for metrics, 1M for good metrics but relatively worse extrapolation. Looking at the periodicity of the wave - the big difference lies in higher dims + higher frequencies.
1
2
37
7,946
Kicked off TSFM tangents today with an amazing lecture from @chengstaa . Spinning up *good* human annotation pipelines is so challenging and what Claire has managed to accomplish at Cohere is nothing short of incredible! Happy to have learned from her
37
1,931
Nothing beats walking around downtown Toronto at 10:30pm listening to Dhoom Machale on repeat
4
1
34
3,490
Sabbatical Micro Report #1:
Time flies and it has somehow been ~4 months since I went on a break to focus on health. Happy to report that I've been doing great and the break feels very rejuvenating!
- granter dot ca : MVP to help NPOs find grants. A few users including YV but not useful enough yet.
- tsfm dot ca: teach people foundation modelling from (near) scratch. We are halfway through the first season with lecture 7 happening this Thursday.
- footnote dot work: Chat and learn like you think. A few paid users with most of my learning happening in Footnote.
Micro Report #2 after season 1 of TSFM has wrapped.
1
36
1,548
> Command-R is the first in a series of model releases advancing capabilities crucial to enterprise adoption at scale.
LFG
⌘-R
Introducing Command-R, a model focused on scalability, RAG, and Tool Use. We've also released the weights for research use, we hope they're useful to the community!
txt.cohere.com/command-r/
2
3
31
2,371
My favourite part of running TSFM is doing outreach, getting ghosted, and not giving up until I get *rejected*
Just say No to me guys, it’s ok
5
33
1,340
Wear Vengeance.
Crafted for the bold. Built to endure.
Pure, heavyweight, garment-dyed cotton.
Design by the one and only @mandiramidha
2
2
34
3,547
Gen-4 is here, go play with it :))
Today we're introducing Gen-4, our new series of state-of-the-art AI models for media generation and world consistency. Gen-4 is a significant step forward for fidelity, dynamic motion and controllability in generative media.
Gen-4 Image-to-Video is rolling out today to all paid plans and Enterprise customers.
1/8
1
2
34
1,592
It was super fun hosting the first C4AI meetup at Cohere HQ in Toronto! Join us at cohere.for.ai :)
2
3
31
More confirmed guest lectures:
- Humans of AI: Annotations for LLMs by @chengstaa from Cohere
- Scaling search space by [REDACTED] from Isomorphic Labs
Only at Toronto School of Foundation Modelling.
1
32
6,871
Seeking sponsorships for Toronto School of Foundation Modelling. If you're interested in supporting, please reach out!
More info here:
Funds - tsfm dot ca / donate
Mission - tsfm dot ca / thesis
1
4
31
3,375
Wrapped the very first demo day of TSFM.
I demoed Footnote.
@akaash gave us a walkthrough of all his transformer experiments. 🐐🐐
Tim talked about his extensive dev tooling customizations.
Patrick talked about Canada’s productivity problem.
1
1
33
2,467
Excellence in Foundation Modelling, cultivated in Toronto.
Made possible with the support of Vengeance, Cohere, New Systems, Modal, Datamarket and Boom Video.
1
4
33
3,327
8 passengers, 4 rides. pretty fun - chatted about AI research, Toronto and its tech/art talent, nature of ambition, the wild concept of Greektown. Discovered new music, learned new things about people I interact with. Might do this on a recurring basis
6
1
29
4,207
Toronto School Of Foundation Modelling acceptance emails have gone out! See you in class in a week :))
2
2
29
3,236
TSFM: Optimizers by @yacinelearning on September 25th will take place at @cohere Toronto office. Shoutout Cohere (again) for all the support
2
3
30
4,051
- The owner at Golden Turtle stopped me on my way out yesterday, gifted me a mug and said “Thanks for being such a good customer”
- Pictured, I got exactly what I ordered from Saffron spice but they wrote me a note saying they didn’t have my usual on the menu that day
- The whole team at Karahi Boys asks me how I’ve been every time I go there.
This is what loving a city looks and feels like. It’s so rewarding, being a recurring side character in all these people’s lives. Best city in the world
2
1
30
1,391
Great day at the Stadium today.
- good focus session in the basement
- casually met the creator of my action button trigger (@superwhisper @neilsuperduper)
- got a copy of “A decade of Kid Studio”
- great conversations with cool people
God Bless Toronto!
2
2
32
1,553
I see so much status chasing in AI startups today. Optimizing for stuff that’s great on paper to maximize the amount you can raise. A company is fundamentally about building, get people that love to get their hands dirty and build. Everything else will follow and if it doesn’t, you tried and that’s really the only thing that matters - that you tried something you genuinely believed in and is personal to you.
1
2
30
1,537
was talking to a friend about how I’d love to see TSFM-like initiatives in other fields but with a strong emphasis on in-person presence, taking you from novice to self-sufficient in 3 months. I paused for a moment after that and had a tech bro discovers school moment
31
957
The prompt coherence of Frames is unlike anything I've seen before. If you have a vision for a shot, take the time to describe it well and Frames will make something that's well worth your time
Today we are releasing Frames. Our most advanced base model for image generation, offering unprecedented stylistic control and visual fidelity. Learn more below.
(1/10)
30
1,126
Stadium logs 29092025
- had a daytime beer for the first time ever with @tommytrinh
- learned about what @MogtabaAlim is building and some crazy stats on emergency response times in GTA
- @bram and I demoed our products to each other
- Got great feedback on Footnote from @lychkel
7
2
30
2,285
A documentary that I helped produce (just a lil bit) is premiering this Monday. Witness Lucy Shepherd do the impossible and explore the uncharted territories of Deep Amazon! If you’re in the UK, catch it on Channel 4 at 10pm. I’m so excited for this :)
2
2
30
1,765
Just got my driver’s license, the world is my oyster. Also, I don’t fw pedestrians and bikers anymore
2
27
2,931
My CS undergrad was way more useful than I realized at the time. I learned to write compilers, work with assembly, and understand CPU/GPU architectures and to program them efficiently. I just wish I’d had a deeper incentive to explore them further back then. It’s no surprise that building a natural language interface between humans and computers requires a deep understanding of both systems. Being in AI dev now gives me the perfect reason to rebuild that holistic mental model.
1
1
29
2,169
When TSFM S01 wraps, we will have a celebration at the Main Sewage Treatment Park in Toronto
1
1
27
1,684
