Today we are excited to announce our collaboration with @nvidia. We are releasing two domain-specific open weight models fine-tuned on Nemotron 3.5 Lightning. Neither was post-trained by a human.
Instead, both models were built using the Fastino Fine-Tuning Agent, our autoresearch agent that handles the full post-training pipeline on its own: from task research and data curation, to eval construction, parallel training experiments, and model deployment.
- Fastino-Nemotron-3.5-Lightning-Finance excels at numerical reasoning, research, and summarization over financial concepts and documents. When evaluated, the fine-tuned model increased FinQA execution accuracy by 43.37 points (from 15.86% to 59.23%) and improved 7.81 points (from 49.65% to 57.46%) on BizFinBench.
- Fastino-Nemotron-3.5-Lightning-Healthcare specializes in clinical conversation quality, summarizing notes and discharge documentation, and extracting medical concepts from unstructured text. In post-training, the Fastino Fine-Tuning Agent improved performance on a variety of healthcare benchmarks, including increasing MEDEC flag accuracy by 11.32 points, putting it ahead of Claude Opus 4.6 and beating Muse Spark by 0.98 points on the MedHELM leaderboard.
Our @fastinoAI research team found that given the same time, compute constraints, and data, Nemotron 3.5 Lightning learns and converges faster than other similarly-sized open source when fine-tuned, while still retaining its original capabilities.
Both models are available on @huggingface under the Apache 2.0 license.
Starting today, Fastino Fine-Tuning Agent is open for private preview. Sign-up link and links to the model weights in the thread. Big shoutout to our lead author on the agent paper @DhruvAtreja1!
@Docusign I have been waiting over a week for a callback for a simple account issue via support.
I have a paid account. This issue is preventing us from using DocuSign for active business transactions and constitutes a production-impacting outage.
Can someone please escalate?
A fantastic first day for our UltraFlow Similarity Search! Just an hour and a half into the trading day, our users were able to predict the shape of the rest of the day on $SPY.
Similarity Search is an indicator that finds similar historical price patterns and displays how the price moved on those days.
#SPY #QQQ #daytradingindicators #tradingindicators #SPYtrades #BigShort #daytrading #daytrader #stockmarket
Net Option Flow looks overall bullish on $GME, however there was a HUGE bearish market maker manipulation attempt at 10am EST. Letโs see if market makers are able to keep prices down or if retail will be able to overwhelm them!
#daytrader#daytrading#daytraders#stockmarket #GME #ICT #wallstreetbets #MOASS
Someone asked for us to post the $GME chart today, looks like momo and smart money were moving in the same direction a lot today, and currently both are selling. Net Option Flow was looking bullish earlier this morning starting from 9:50am EST, but has now tapered off. Weโll see how the day shapes up, we have new data coming in every moment on our real-time charts!
To learn how to read our charts, check out this blog post: https://t.co/6Wgk5JKEAh
#daytrading #daytrader #daytraders #StockMarket #ICT #gme #wallstreetbets
We're excited to announce our new beta feature from BigShort - Net Option Flow! It's a groundbreaking way to look at options volume, and the user reception so far has been fantastically positive.
https://t.co/VPjpdoFh3s
Check out our members-profits channel in our discord, it's open and free for anyone to join, our members have been kind enough to share their trades there!
#daytrading #StockMarket #ICT #daytrade #BigShort
$ABNB jumped in price today, and our users knew it was coming! SmartFlow (green line) and MomoFlow (red line) showed a clear divergence, with SmartFlow moving up as smart money bought, while at the same time, momo sold. This upward move was further confirmed by increasingly bullish Manipulation, as market makers suppressed prices while they accumulated shares, before letting the price fly.
Another great trade from BigShort! #DayTrading #StockMarket
Maybe JPM actually looked to see how $PYPL branded checkout share is stable w/volume, revenue & engagement growth all accelerating.
Or how its branded & unbranded flows deliver large checkout edges in availability, authorization & chargebacks vs. the rest of the industry.
Or how the mere presence of its button raises shopper conversion by 8% vs. 6% previously thanks to its brand strength & buyer protection program.
Or how $AAPL sees the value in more deeply partnering w/the giant 2-sided network just like $V & $MA did a decade ago.
Or how global smartphone market fragmentation makes $PYPL an ideally agnostic layer of unification.
Or how Braintree share gains are robust w/margin levers as clear as day & a new CEO w/direct experience in pulling those levers.
Or how Venmo ARPU is in inning 1 w/checkout, cards & business profiles all ramping.
Or how a company doing well over $1 trillion in annual volume may not suddenly die.
Or how meaningful of a buyback program it has in place w/the cash flows & the balance sheet to easily support the repurchases.
& how maybe a firm with all of these factors & a 15% EPS CAGR shouldn't trade for 12x 2023 earnings.
Just maybe. ๐คทโโ๏ธ๐
6) Acquired Drizly ($1.1B), Cornershop ($1.4B), Postmates ($2.65B). Growth with profit.
You can be lean, aggressive and keep taking bold bets. Always play to win. #startups#venturecapital#profitability
5) Grew Uber One members to 12M, up 2x YoY. Focus on lifetime value over margins.
Uber One offers discounts and free delivery. Lower margin per order, but higher retention and order frequency. - foundation towards profitability. A card from Amazon's sleeve.