Rohit Goswami

4K posts

Rohit Goswami banner
Rohit Goswami

Rohit Goswami

@rg0swami

Doctoral Researcher @ UI. Software Engineer, @quansightai Labs. FOSS. ex-Senior Project Associate @ IITK. Code, coffee, academia, @[email protected]

University of Iceland Katılım Ocak 2019
3.4K Takip Edilen353 Takipçiler
Rohit Goswami retweetledi
Fortran
Fortran@fortranlang·
There's still time to submit an abstract to FortranCon 2025. What are you waiting for? events.fortrancon.org/event/1/ Final submission deadline is September 12.
English
0
11
32
3.1K
Rohit Goswami
Rohit Goswami@rg0swami·
@costa_p :'D depending on the sophistication of the "code" in question a lot of them emit NaNs and "soldier on" (especially for machine learning)
English
0
0
0
39
Pedro Costa
Pedro Costa@costa_p·
Unsolicited advice, not a bad idea to add this 👇 when running on shitty machines. Welcome! #fml 😡
Pedro Costa tweet media
English
2
0
9
697
Sterling G. Baird
Sterling G. Baird@SterlingBaird1·
Who would be interested in participating in a hackathon for Bayesian optimization focused on chemistry and materials science applications?
English
48
8
134
28.8K
Rohit Goswami
Rohit Goswami@rg0swami·
@SmitBerend @kmjablonka Isn't the real problem trusting the results on unknown / new data? The fact that the LLMs can very confidently predict rubbish very easily didn't seem to be covered in any detail..
English
0
0
0
21
Berend Smit
Berend Smit@SmitBerend·
Large-language models will change #CHEMISTRY. See 15+ examples of answers to chemical questions. What is common to those: @kmjablonka et al have near zero knowledge but we could fine-tune #GPT3 to get models that do better than the state of the art. rdcu.be/dxZ3x
Berend Smit tweet media
English
1
10
61
5.1K
Rohit Goswami
Rohit Goswami@rg0swami·
This is an OpenPGP proof that connects my OpenPGP key to this Twitter (X) account. For details check out keyoxide.org/guides/openpgp… my OpenPGP key: openpgp4fpr:74b1f67d8e43a94a755407689ccce36402cb49a6]
English
0
0
0
43
Rohit Goswami
Rohit Goswami@rg0swami·
@thegautamkamath This is pretty toxic and sets them up for failure. Research takes time, and letting students feel like a set problem to be solved is the same as a fully fleged paper, is a misake. Undergraduates could and should focus on engineering skills instead of working "on a paper".
English
0
0
2
29
Rohit Goswami retweetledi
Trey Hunner
Trey Hunner@treyhunner·
Python 3.7 reached the end of its life today. 🎉 Here are some Python 3.8 features you can now embrace in your open source code: • cached_property • self-documenting f-string expressions • walrus operator • positional-only args • passing dictionaries to reversed() #Python
English
2
44
186
22.5K
Rohit Goswami
Rohit Goswami@rg0swami·
@ProfMaxNew Undergraduate overconfidence. As part of most curricula people write some half-baked parser to a toy language and imagine its all done and dusted.
English
0
0
1
16
Max ⊶ New @maxsnew@types.pl
Max ⊶ New @[email protected]@maxstewartnew·
When did we as a CS community decide that parsing was a "solved" problem? It's a very common sentiment so I wonder if it originates from a source we can actually identify
English
20
4
67
17.8K
Rohit Goswami retweetledi
Debabrata Goswami
Debabrata Goswami@Debabra59832405·
Great news! My article, "Quality assessment of the commercially available alcohol-based hand sanitizers with femtosecond thermal lens spectroscopy" has just been accepted for publication in the Open Access journal @PeerJAnalytical
English
1
1
4
242