Adam Bloebaum
@bl0ebaum
Followers
377
Following
3K
Media
19
Statuses
210
Engineer @DrivelineBB | Pitcher @Nationals | Alum @UW | Student @GeorgiaTech
Seattle, WA
Joined March 2019
RT @hedgertronic: Had a great mid-season meeting with the Phillies player development staff that made me reflect on how many low-future-val….
0
19
0
RT @hedgertronic: After hours of research (playing every day), here is my proposed solution to one of the great unsolved problems in pre-ga….
0
6
0
RT @jacklambert__: Best ERA (min 20 IP) across all levels this year:. Cole Winn (AAA/MLB): 0 ERA, 35 IP.Tony Rossi (A/A+): 0 ERA, 21.1 IP.C….
0
6
0
RT @Crider_HP: “The highest technique is to have no technique. My technique is a result of your technique; my movement is a result of your….
0
6
0
RT @SamEhrlich: With the NCAA baseball tournament starting today, I built an Elo rating model to simulate the College World Series. Here's….
0
6
0
RT @WilmBlueRocks: The new guy with the win❗️. In his first Blue Rocks outing, @bl0ebaum records a 2 inning no-hit extra-innings win 🔥. 2 I….
0
6
0
RT @milb_central: The Washington Nationals are promoting Adam Bloebaum to the Wilmington Blue Rocks (A+), per @_chaseford. .
0
9
0
RT @Tstokey: Full signal attack angle visual from the Biomechanics assessment in the Launchpad @DrivelineBB . CV Bat Tracking 🤌🤌🤌 https://t….
0
5
0
RT @JackSmithPxP: "Just throw more. Throw different things.". Of all 30 Carolina League pitchers with a 0 ERA, none have logged as many inn….
0
5
0
RT @Nats_PlayerDev: .@bl0ebaum hasn't allowed an earned run all season, tossing 2.0 more scoreless innings last night!. 17.2 IP.24 K.0.00 E….
0
14
0
RT @JoeDoyleMiLB: Between Aiva Arquette and Youngstown State OF Kyle Fossum, the University of Washington sure did lose a ton of offense fr….
0
14
0
RT @Crider_HP: Best in class tools. Best in class training. Built by scumbags who definitely want it as bad as you do or more. Come holl….
0
4
0
Put my physics degree to good use and wrote about tendon adaptations for throwers—what a tendon is, why it matters for throwing, how it adapts, and how to actually train with that in mind. Give it a read, or don’t—but it’s #good2learn.
0
6
33
RT @Crider_HP: Efficient, high output plyometric contacts require both a strong and well timed muscular contraction and a stiff, compliant….
0
1
0
Moondream shines with its tiny footprint, instant CPU/edge inference & simple prompts for multiple tasks. A lightweight, versatile, budget‑friendly CV model with endless possibilities.
0
0
1
The language model is a Phi‑variant quantized to INT4 (~1 GB footprint) and fine���tuned on visual prompts for captioning, VQA, OCR, pointing & more.
1
0
0
The cross‑modal projector uses a lightweight Q‑Former to transform 1,152‑dim vision embeddings into a sequence of pseudo‑tokens that the language model can attend to alongside text, enabling unified vision+language reasoning.
1
0
0
The vision encoder uses a Patch‑ViT backbone pretrained with SigLIP on web‑scale image–text pairs. It splits images into 14×14 patches (588‑dim), projects them to 1,152‑dim, then runs them through 12 fp16 transformer blocks to capture objects, relationships & context.
1
0
0