Al Merose

@al.merose.com

3918 Followers 2108 Following

Differential database administrator. Machine learning for climate & weather. Ex-Google Research, Founding Member of Technical Staff @OpenAthena.ai. Tinkering with xql.systems. He/him.

I checked - and I have to be straight with you: it's not recoverable from the bucket.
16. * Show HN: I implemented a neural network in SQL (github.com/xqlsystems) 9 points by alxmrs 35 minutes ago | hide | discuss | edit
An unbelievably cute grey tabby cat with her eyes closed resting her head on her paws in a basket like thing.
% uv run benchmarks/nn.py warning: The Python request from `.python-version` resolved to Python 3.11.11, which is incompatible with the script's Python requirement: `>=3.12` Installed 41 packages in 232ms step 0: loss 2.298 train_acc 0.154 test_acc 0.145 step 5: loss 1.559 train_acc 0.440 test_acc 0.325 step 10: loss 1.150 train_acc 0.642 test_acc 0.610 step 15: loss 0.865 train_acc 0.702 test_acc 0.620 step 20: loss 0.825 train_acc 0.716 test_acc 0.665 step 25: loss 0.686 train_acc 0.752 test_acc 0.750 step 30: loss 0.598 train_acc 0.800 test_acc 0.760 step 35: loss 0.584 train_acc 0.788 test_acc 0.720 step 40: loss 0.484 train_acc 0.864 test_acc 0.815 step 45: loss 0.582 train_acc 0.792 test_acc 0.735 step 50: loss 0.504 train_acc 0.800 test_acc 0.775 step 55: loss 0.452 train_acc 0.830 test_acc 0.730 step 59: loss 0.413 train_acc 0.864 test_acc 0.785 trained (784, 196, 32, 10) MLP; weights -> xarray {'inp_0': 785, 'out_0': 196, 'inp_1': 197, 'out_1': 32, 'inp_2': 33, 'out_2': 10}. <xarray.Dataset> Size: 1MB Dimensions: (inp_0: 785, out_0: 196, inp_1: 197, out_1: 32, inp_2: 33, out_2: 10) Coordinates: * inp_0 (inp_0) int64 6kB 2 3 4 5 6 7 8 9 ... 778 779 780 781 782 783 784 * out_0 (out_0) int64 2kB 9 13 20 27 33 45 48 ... 155 166 168 181 187 189 * inp_1 (inp_1) int64 2kB 0 1 2 3 4 5 6 7 ... 190 191 192 193 194 195 196 * out_1 (out_1) int64 256B 0 2 4 6 8 22 25 28 29 ... 7 16 19 24 3 10 18 21 * inp_2 (inp_2) int64 264B 0 1 2 3 4 5 6 7 8 ... 9 13 20 27 12 14 30 31 32 * out_2 (out_2) int64 80B 1 0 2 4 6 8 3 5 7 9 Data variables: layer_0 (inp_0, out_0) float64 1MB -0.05519 0.05624 ... -0.04738 0.01336 layer_1 (inp_1, out_1) float64 50kB 0.1518 -0.1549 ... 0.02047 -0.05072 layer_2 (inp_2, out_2) float64 3kB -0.04275 0.1687 ... -0.04326 -0.05491
WITH RECURSIVE params (step, a, b) AS ( SELECT 0, 0.0, 0.0 UNION ALL SELECT params. step + 1, params.a - 1r*AVG (grad(loss, a)), params.b - 1r*AVG (grad(loss, b)) FROM params CROSS JOIN d WHERE params.step ‹ STEPS GROUP BY params.step, params.a, params.b) SELECT * FROM params ORDER BY step
A screenshot. Text: YOU ARE... The True Believer patron saint: Amanda Askell AI is real, it's powerful, and handled with genuine care it can be one of the best things we ever build. You think hard about its character, what it's like, what we owe it and it owes us. You're optimistic the way a thoughtful person is optimistic — clear-eyed, not starry-eyed.
Mythos on Paxos
SELECT r.region, AVG(a."2m_temperature") - 273.15 AS avg_c FROM era5.surface a JOIN regions r ON a.latitude BETWEEN r.lat_min AND r.lat_max AND a.longitude BETWEEN r.lon_min AND r.lon_max WHERE a.time BETWEEN TIMESTAMP '2020-06-01' AND TIMESTAMP '2020-06-01 23:00:00' GROUP BY r.region
SELECT f.model, f.prediction_timedelta AS lead, SQRT(AVG(POWER(f."2m_temperature" - e."2m_temperature", 2))) AS rmse FROM forecasts f JOIN era5 e ON e.time = f.time + f.prediction_timedelta -- valid_time = init + lead AND e.latitude = f.latitude AND e.longitude = f.longitude GROUP BY f.model, f.prediction_timedelta
Twiggy with her eyes closed on my couch.
Mythos is a brand of Greek beer featuring a unicorn.
Calvin: HOW DO THEY KNOW THE LOAD LIMIT ON BRIDGES, DAD? Dad: (linear systems of equations representing the physics of bridge load) Calvin: OH. I SHOULD'VE GUESSED.
Routed Expert Normalization and Rescaling Marin is open-development: anyone can follow experiments and contribute. Elie Bakouch, a community member, recently noticed a difference between our expert weighting and DeepSeek's*. Within hours of his suggestion to renormalize and scale the routed expert outputs, we confirmed the small boost and added it to the recipe. Without his recommendation, this improvement would not have made it into our next large scale run. When evaluating the change, we compared to the in-progress recipe indicated by the blue dots below.
My grey tabby cat sitting over a vibrant painting of an artist’s interpretation of jazz. Painting are on the wall behind her.
import xarray as xr import xarray_sql as xql # Open a year of ARCO-ERA5 — all 273 variables. Selecting a year up front # keeps Dask's partition setup cheap before any chunks are read from GCS. ds = ( xr.open_zarr('gs://gcp-public-data-arco-era5/ar/full_37-1h-0p25deg-chunk-1.zarr-v3', chunks=None, storage_options={'token': 'anon'}) # Anonymous read from the public GCS bucket — no auth required. .sel(time='2020') .chunk({'time': 1}) ) ctx = xql.XarrayContext() ctx.from_dataset('era5', ds, table_names={ ('time', 'latitude', 'longitude'): 'surface', ('time', 'level', 'latitude', 'longitude'): 'atmosphere', }) # Registration: ~0.5s for a full year of hourly ERA5, all variables. # Heads up: ARCO-ERA5 has 262 surface + 11 atmospheric variables. The library # pushes column projection down to Zarr, so SELECT only fetches what you ask # for — but `SELECT * FROM era5.surface` would try to pull every variable # across the year (terabytes from GCS). # ---> Always SELECT specific columns. <--- # Average 2m-temperature over NYC on the morning of 2020-01-01. The library # pushes WHERE clauses on dimension columns down to partition pruning. ctx.sql(''' SELECT AVG("2m_temperature") - 273.15 AS avg_c FROM era5.surface WHERE time BETWEEN TIMESTAMP '2020-01-01' AND TIMESTAMP '2020-01-01 05:00:00' AND latitude BETWEEN 39 AND 40 AND longitude BETWEEN 286 AND 287 -- ERA5 uses 0-360 longitudes ''').to_pandas() # avg_c # 0 8.640069 # Average temperature per pressure level, globally. ctx.sql(''' SELECT level, AVG(temperature) - 273.15 AS avg_c FROM era5.atmosphere WHERE time BETWEEN TIMESTAMP '2020-01-01' AND TIMESTAMP '2020-01-01 05:00:00' GROUP BY level ORDER BY level DESC ''').to_pandas()
Everything is on autopilot for the night. Is there anything else you'd like to work on, or shall we wrap up this session?
4 Experiments We now apply our method to analyze how embedding spaces evolve during LLM-related learning tasks. Unless stated otherwise, all results pertain to the model's final hidden layer, i.e., T_ı as defined in Section"3. We refer the reader to Section"C.2 for results on layers other than the last. Subsequently, we will focus on four central questions: (Q1) How does fine-tuning on different datasets al- ter latent space geometry? (Q2) How can local dimension estimates detect grokking? (Q3) How can local dimension estimates detect the limit of training capabilities? (Q4) How can local dimension estimates detect overfitting?
ANUPONG CHANTORN Animal-Man Family Bronze, 2010 I have taken the idea of the realm of the hungry ghosts (peta), a cautionary tale which warn us to fear the law of karma, to not commit sinful acts, and reminds us to live morally, and use it as an inspiration for the "Animal-Man Family" sculptures. The decline of the modern Thai family institution and its lack of morality is reflected in the animal-human hybridized form, which is created by the mind, as it determines the manifestation of the result of karma, and beast-like behaviours.
ANUPONG CHANTORN Animal-Man Family Bronze, 2010 Bangkok Contemporary Art Museum
(Mouse from the matrix bursting into the cantina): “Ox/EC Found an error in ERA5 using GraphCast”
The last latke from our batch, with sour cream and a heaping pile of dill. On kind of a messy plate.
A picture of the book by ARTHUR GILL: “MACHINE AND ASSEMBLY LANGUAGE PROGRAMMING OF THE PDP-II”
A sign that says “Los Alamos Perch, est 2017”. On top of the antique shop is a lit up cross and star of David (the six pointed star is above the cross).
An otter in the Monterey Bay chowing down on sea urchin.
Code demoing Xarray-SQL. It reads: ``` import xarray as xr import xarray_sql as xql ds = xr.tutorial.open_dataset('air_temperature') # The same as a dask-sql Context; i.e. an Apache DataFusion Context. ctx = xql.XarrayContext() ctx.from_dataset('air', ds, chunks=dict(time=24)) # the dataset needs to be chunked! result = ctx.sql(''' SELECT "lat", "lon", AVG("air") as air_total FROM "air" GROUP BY "lat", "lon" ''') ```
Figure 2 Global distribution of data for OlmoEarth pretraining. We randomly sample 285,288 locations based on OpenStreetMap categories. What's your favorite map projection? I like the Peirce quincuncial projection centered on Antarctica. They said it didn't make any sense for this figure though. They said "We hate Antarctica, that's why we don't sample any points there". I said, "Hey, there is one point there!" They didn't have a response to that so they tried to silenced me by deactivating my acc
Puns are not the games of a mediocre mind. Duchamp found in them "a source of stimulation both because of their actual sound and because of unexpected meanings attached to the interrelationships of disparate words [...]. Sometimes four or five different levels of meaning come through. If you introduce a familiar word into an alien atmosphere, you have something comparable to distortion . in painting, something surprising and new."42
The Ubuntu 24 boot screen, which my reflection in the black of the screen
Sandra Mujinga's practice… is concerned with how we relate to bodies that are "other". Mujinga stages encounters with beings unknown to us. The bodies that populate her worlds often seem to be technological in nature, transcending human form. …These are ways of rendering a body invisible or unrecognizable-and, in the case of multiplication, of providing kinship-and thereby offering protection. Her interest in such survival strategies may be linked to broader concerns about how Black bodies are watched and perceived-how they are often subjected to constant surveillance yet rarely truly acknowledged or represented in public spaces or positions of power. For Skin to Skin, Mujinga has transformed the museum's lower-level exhibition space into an otherworldly universe inhabited by 55 figures. Drenched in green light and immersed within a soundscape composed by the artist herself, the cloth-wrapped forms stand among mirrored pedestals that multiply their presence.
Close up: black pen on white paper, the face if a woman over and over spilling into each other.
On the DOGE for AI website, a slide that reads: Will DOGE 1.0 have impacts in other areas of our civilization? Picture it: In HR disputes, employees who file bully reports can be terminated (teamblind.com). What if we have Al-arbitrators? In insurance claims, what if Al-arbitrators stand between policyholders / adjusters? In IRS tax audits, what if Al-arbitrators stand between taxpayers / examiners? HOH Al arbitrators excel at logic-based and fact-checking disputes to ensure justice ... Can we think bigger? International trade disputes? Environmental disputes? Geopolitical conflicts?
A sticker on a blue Subaru that reads “Aromas Pace Car”. Below the words are a row of snails. The sticker is a lighter shade of blue than the car.
Jason S. Campbell tweets: ``` Charlie Kirk: "It's worth to have a cost of, unfortunately, some gun deaths every single year so that we can have the Second Amendment" ```
Album cover: stereolab’s “dots and loops”
Album Cover for of Montreal’s Cherry Peel
But permutation invariance means that the Perceiver's architecture cannot in and of itself exploit spatial relationships in the input data. **Spatial relationships are essential for sensory reasoning (Kant, 1781)** and this limitation is clearly unsatisfying. In the attention literature, position information is typically injected by tagging position encodings onto the input features (Vaswani et al., 2017); we pursue this strategy here as well. While position information is typically used to encode sequence position in the context of language, it can also be used to encode spatial, temporal, and modality identity.
A grey tabby cat pounces on my cell phone as it plays the julia set animation that I am quote skeeting.
A google flights search results page. Next to the flight, there’s a warning that says: “Contrail warming potential: Medium” Then, when you hover over it, there is a dialog that says: Contrails, or "Condensation trails," are the line-shaped clouds behind planes and contribute about a third of aviations total climate impact. Learn more about contrails.
Trump, Claiming Weak Jobs Numbers Were 'Rigged, Fires Labor Official Economists said President Trump's decision to oust the head of the Bureau of Labor Statistics could undermine confidence in government economic data.
A screenshot of the Spotify app of the audiobook. The text reads: “Included in Premium Anti-Oedipus: Capitalism and Schizophrenia Gilles Deleuze • Felix Guattari + 2 others # 4.8 (10) • Dec 4, 2023 • 21h 26m About Chapters More like this Try out the book Listen to samples, clips, and more.”
Screenshot of my audioboom that reads “Chapter 2: Worthy and Unworthy…” of “Manufacturing Consent: The …”
A picture of a half completed tub storage system. Its a wide, rectangular wooden structure where you can slide in tubes. I plan to put it in my garage.
Epic handshake meme Left: wine culture Right: cocktail culture Handshake: vermouth is where it’s at
I’m in a yellow button down shirt holding ol’ toothache, an antique accordion that was my wife’s grandfather’s. It reads “M. HOHNER” on the side. From my finger position, a musician would be able to tell that I don’t know what I’m doing.
Joan Westenberg on LinkedIn saying: “If I see one more post about emdashes or how to identify if a post has been written by Al, I swear to God I will turn Linkedin around and go straight home.”
The Google Earth Engine code editor displaying an evaporation map loaded directly from ARCO-ERA5! The black and white map overlooks the continental US. It's easy to make out land masses, but there are also cloud-looking visuals on the map.
Umberto Eco, Travels in Hyperreality. Essay: “Dreaming of the Middle Ages”, 1986. Text reads: 4. The Middle Ages of Romanticism, with their stormy castles and their ghosts. Germane to the eastern cruelty of Vathek, these Middle Ages return in some contemporary space-operas, where it is enough to put computers in the dungeon to transform it into a starship.
A gold metal Timex watch that reads, in an analog face, 12:42:44, on my left wrist.