You are nothing but useless leeches in the Internet age.
You must go bankrupt all of you, ASAP.
Research paid with taxpayer money must be made available for free.
Researchers and reviewers all work for peanuts, while academic publishers get money for doing the work that an algorithm could do. OurBigBook.com.
When Ciro learned URLs such as www.nature.com/articles/181662a0 log you in automatically by IP, his mind blew! The level of institutionalization of this theft is off the charts! The institutionalization of theft is also clear from article prices, e.g. 32 dollars for a 5 page article.
Long live the Guerilla Open Access Manifesto by Aaron Swartz (2008).
Key physics papers from the 50's are still copyright encumbered as of 2020, see e.g. Lamb-Retherford experiment. Authors and reviewers got nothing for it. Something is wrong.
Infinite list of other people:
- blog.machinezoo.com/public-domain-theft by Robert Važan:
Scientific journals are perhaps one of the most damaging IP rackets. Scientists are funded by governments to do research and publish papers. Reviews of these papers are done by other publicly funded scientists. Even paper selection and formatting for publication is done by scientists. So what do journals actually do? Nearly nothing.
It is hard for complex organisms to evolve because longer DNA means longer replication time Updated 2025-01-03 +Created 1970-01-01
Because DNA replication is a key limiting factor of bacterial replication time, such organisms are therefore strongly incentivized to have very minimal DNAs.
Power, Sex, Suicide by Nick Lane (2006) 7 "Why bacteria are simple" page 169 puts this nicely:
Bacteria replicate at colossal speed. [...] In two days, the mass of exponentially doubling E. coli would be 2664 times larger than the mass of the Earth.Luckily this does not happen, and the reason is that bacteria are normally half starved. They swiftly consume all available food, whereupon their growth is limited once again by the lack of nutrients. Most bacteria spend most of their lives in stasis, waiting for a meal. Nonetheless, the speed at which bacteria do mobilize themselves to replicate upon feeding illustrates the overwhelming strength of the selection pressures at work.
github.com/CovertLab/WholeCellEcoliRelease is a whole cell simulation model created by Covert Lab and other collaborators.
The project is written in Python, hurray!
But according to te README, it seems to be the use a code drop model with on-request access to master. Ciro Santilli asked at rationale on GitHub discussion, and they confirmed as expected that it is to:
- to prevent their publication ideas from being stolen. Who would steal publication ideas with public proof in an issue tracker without crediting original authors? Academia is broken. Academia should be the most open form of knowledge sharing. But instead we get this silly competition for publication points.
- to prevent noise from non-collaborators. But they only get like 2 issues as year on such a meganiche subject... Did you know that you can ignore people, and even block them if they are particularly annoying? Much more likely is that no one will every hear about your project and that it will die with its last graduate student slave.
The project is a followup to the earlier M. genitalium whole cell model by Covert lab which modelled Mycoplasma genitalium. E. Coli has 8x more genes (500 vs 4k), but it the undisputed bacterial model organism and as such has been studied much more thoroughly. It also reproduces faster than Mycoplasma (20 minutes vs a few hours), which is a huge advantages for validation/exploratory experiments.
The project has a partial dependency on the proprietary optimization software CPLEX which is freeware, for students, not sure what it is used for exactly, from the comment in the
requirements.txt the dependency is only partial.This project makes Ciro Santilli think of the E. Coli as an optimization problem. Given such external nutrient/temperature condition, which DNA sequence makes the cell grow the fastest? Balancing metabolites feels like designing a Factorio speedrun.
There is one major thing missing thing in the current model: promoters/transcription factor interactions are not modelled due to lack/low quality of experimental data: github.com/CovertLab/WholeCellEcoliRelease/issues/21. They just have a magic direct "transcription factor to gene" relationship, encoded at reconstruction/ecoli/flat/foldChanges.tsv in terms of type "if this is present, such protein is expressed 10x more". Transcription units are not implemented at all it appears.
Everything in this section refers to version 7e4cc9e57de76752df0f4e32eca95fb653ea64e4, the code drop from November 2020, and was tested on Ubuntu 21.04 with a docker install of
docker.pkg.github.com/covertlab/wholecellecolirelease/wcm-full with image id 502c3e604265, unless otherwise noted.Array of
Elf64_Shdr structs.Each entry contains metadata about a given section.
e_shoff of the ELF header gives the starting position, 0x40 here.e_shentsize and e_shnum from the ELF header say that we have 7 entries, each 0x40 bytes long.So the table takes bytes from 0x40 to
0x40 + 7 + 0x40 - 1 = 0x1FF.Some section names are reserved for certain section types: www.sco.com/developers/gabi/2003-12-17/ch4.sheader.html#special_sections e.g.
.text requires a SHT_PROGBITS type and SHF_ALLOC + SHF_EXECINSTRRunning:outputs:
readelf -S hello_world.oThere are 7 section headers, starting at offset 0x40:
Section Headers:
[Nr] Name Type Address Offset
Size EntSize Flags Link Info Align
[ 0] NULL 0000000000000000 00000000
0000000000000000 0000000000000000 0 0 0
[ 1] .data PROGBITS 0000000000000000 00000200
000000000000000d 0000000000000000 WA 0 0 4
[ 2] .text PROGBITS 0000000000000000 00000210
0000000000000027 0000000000000000 AX 0 0 16
[ 3] .shstrtab STRTAB 0000000000000000 00000240
0000000000000032 0000000000000000 0 0 1
[ 4] .symtab SYMTAB 0000000000000000 00000280
00000000000000a8 0000000000000018 5 6 4
[ 5] .strtab STRTAB 0000000000000000 00000330
0000000000000034 0000000000000000 0 0 1
[ 6] .rela.text RELA 0000000000000000 00000370
0000000000000018 0000000000000018 4 2 4
Key to Flags:
W (write), A (alloc), X (execute), M (merge), S (strings), l (large)
I (info), L (link order), G (group), T (TLS), E (exclude), x (unknown)
O (extra OS processing required) o (OS specific), p (processor specific)The
struct represented by each entry is:typedef struct {
Elf64_Word sh_name;
Elf64_Word sh_type;
Elf64_Xword sh_flags;
Elf64_Addr sh_addr;
Elf64_Off sh_offset;
Elf64_Xword sh_size;
Elf64_Word sh_link;
Elf64_Word sh_info;
Elf64_Xword sh_addralign;
Elf64_Xword sh_entsize;
} Elf64_Shdr;This kind of died at some point checked as of 2023.
Does this contain any structured data? E.g. can you list all papers by a given author besides just searching and hoping there are no homonyms?
mlcommons.org/en/ Their homepage is not amazingly organized, but it does the job.
Benchmark focused on deep learning. It has two parts:Furthermore, a specific network model is specified for each benchmark in the closed category: so it goes beyond just specifying the dataset.
Results can be seen e.g. at:
- training: mlcommons.org/en/training-normal-21/
- inference: mlcommons.org/en/inference-datacenter-21/
And there are also separate repositories for each:
E.g. on mlcommons.org/en/training-normal-21/ we can see what the the benchmarks are:
| Dataset | Model |
|---|---|
| ImageNet | ResNet |
| KiTS19 | 3D U-Net |
| OpenImages | RetinaNet |
| COCO dataset | Mask R-CNN |
| LibriSpeech | RNN-T |
| Wikipedia | BERT |
| 1TB Clickthrough | DLRM |
| Go | MiniGo |
The most important thing this project provides appears to be the
.onnx file format, which represents ANN models, pre-trained or not.Deep learning frameworks can then output such
.onnx files for interchangeability and serialization.Some examples:
- activatedgeek/LeNet-5 produces a trained
.onnxfrom PyTorch - MLperf v2.1 ResNet can use
.onnxas a pre-trained model
There are unlisted articles, also show them or only show them.