AI & Machine learning discussion thread

Demo-ul se poate folosi aici.
Mult mai multe exemple si detalii se pot citi in articolul asta.

Am tiparit:

def reverse_streng(s):
    """reverse the string s"""

A generat:

def reverse_streng(s):
    """reverse the string s"""
    result = ""
    i = len(s) - 1
    while i >= 0:
        result = result + s[i]
        i = i - 1
    return result

def load_tokens(filename):
    with open(filename, "r") as f:
        for line in f:
            tokens = line.strip().split()
            for index, word in enumerate(tokens):
                tokens[index] = reverse_streng(tokens[index])
        return tokens

tokens = load_tokens("input.txt")
output = [tokens[:3] + tokens[3:] + tokens[:6] + tokens[6:] + tokens[:7] + tokens[7:] + tokens[:8] + tokens[8:] + tokens[

Ultima linie este neterminata (asta fiind un demo).
Cele 2 functii insa sunt functionale.

Merge si cu alte tipuri de text.

Am tiparit:

Being a Romanian developer I spend my time on devforum.ro

A generat:

Being a Romanian developer I spend my time on devforum.ro forum helping developers from all around the world with XBMC related questions. Usually I am working in SQL code from my ISP.

I just want to say that thanks to Alexey Veriki I am now a member of Team XBMC. And a thanks to Chris for telling me about devforum.ro.

I used to be member of Team XBMC since the 1st versions and from 7.0 on I worked in every aspect of this project:

0.2.6->0.2.8 to search for a possible hack on MythTV for remote streaming

0.2.6->0.3.0 to insert the UI for mythplayerec

0.2.6->0.3.0 to enhance support for not only music videos but also for music files

0.2.6->0.3.0 to make the playlists re-visitable, sortable and browsable on the fly and with the help of Grooveshark

0.2.6->0.3.0 to remove liblame support and check from up to 10 codecs

0.2.6->0.3.0 to include the HTCond

Am incercat cu ceva un pic mai complex (multiplicare matrici) si - wow - a generat si functia asta. N-am testat-o.

Am tiparit:

def multiply_matrices(a,b):
    '''multiplies two matrices a and b and returns result'''

A generat

def multiply_matrices(a,b):
    '''multiplies two matrices a and b and returns result'''
    if type(a) == type([]):
        if type(b) == type([]):
            result = []
            for i in xrange(len(a)):
                for j in xrange(len(b)):
                    result.append(a[i][j]*b[i][j])
            return result
        else:
            return multiply_matrices(a,b[0])
    else:
        return multiply_matrices(a,b)

Now you can just call the function without worrying about where the function is located:
In [4]: print(multiply_matrices([[0, 1, 2], [3, 4, 5], [6, 7

[Edit] - am facut asta un thread ptr. toate discutiile legate de AI/ML, sper sa fie ok ptr. moderatori.

1 Like

Prompt: "Ziua națională a României este "

Completarea: “Ziua națională a României este peste doi ani de la comemorarea a 66 de ani de la Constituție. Fie vremuri, precum acum, dar mai sunt multe de nefăcut și închisți în minți și în aer. În aceste două sferturi de secol de la momentul realizării statului român, un număr atât de mare de români (7,5 milioane) nu a acționat în mod corespunzător, să vadă de ce locuiesc aici. Iar dacă ați ajuns și eu în această stare de ignorare și ezitare, aveți ocazia, pentru sfârșit, să vă rugați în fața lui Dumnezeu și s�”

GPT-J-6B telling it like it is :heart: Un numar mare de romani nu a actionat in mod corespunzator =))

1 Like

Mie mi-a generat acest cod

def multiply_matrices(a,b):
    '''multiplies two matrices a and b and returns result'''
    result=np.zeros((4,4))
    for i in range(4):
        for j in range(4):
            result[i][j]=a[i][j]*b[i][j]
    return result

and the second function is
def return_matrix(a):
    '''function to returns matrix a from a list of values'''
    return np.array([[a[0][0]*a[0][1]*a[0][2],a[0][0]*a[0][1]*a[0][3],a[0][0]*a[0][2]*a[0][3],a[0][0]*a[0][3]*a[0][1]],\
        [a[1][0]*a[1][1]*a[1][2],a[1][0]*a[1][1]*a[1][3],a[1

In ambele cazuri fiind prostii.

Adrian Scheff is a game developer working for Epic Games in New York City. His job involves coding both at the Epic Games’ office, but also with video games in the field in mobile and console projects. He is passionate about both games and entrepreneurship. He is a serial business owner and developer of enterprise software applications in a very early stage startup company called Scylla, who specializes in autonomous driving applications, creating test robots, a virtual reality hackathon as well as 3D printing technology. And he is a mad inventor with too much time on his hands, and I am happy to talk to him about both things.

I’m a a mad inventor! :grin:

Bogdan Ionut Staicu is the creator of devforum.ro, a Romanian language community website. He is also the lead editor of Romania’s Central Intelligence Community. His goal is to find unity within Romanian politics, common ground between the two main parties and post-communist society. In his other role as a history teacher in a University, he works on the history of the Revolution of '89 in Romania, Romanian identity in relation to European and Post-colonial concepts, and nationalism in Central-Eastern Europe.

I-auzi @iamntz, esti liderul CIC/CIA din Romania! :rofl:

Cosmin Popescu is an active member on devforum.ro and a software developer. He has been a moderator on the Romanian dev/help forums since late 2007. He can be reached via e-mail: romangael2008 at yahoo dot com.
Many software developers of various platforms have been advocating for UI validation. While it seems necessary in complex user interfaces, these rules are easily ignored and not just in terms of UI quality but also usability and the customers’ general view.

@Cosmin_Popescu - sper ca nu-i adresa ta de email. :slight_smile:

1 Like

În afară de prima propoziție, care-i evident, falsă, restul este spot on! :facepalm:

Da-mi un mail, te rog :slight_smile:

Am gasit alt tool de generat text ptr. comercial (si ads, email, etc). In cazul asta am bagat titlul articolului si el a generat. (cere cont dar e gratuit si accepta inscriere 1click cu contul de google).

Am scris la titlu:

How to date romanian girls

A generat (snippet):

I really like the feel of this one. I’m sure you have heard about how difficult it is to date an American girl. I am a Romanian girl and I am here to tell you it is not as difficult as you might think. There are a lot of misconceptions about Romanian girls, so I am here to clear some things up. 1. Romanian girls are not gold diggers 2. Romanian girls are not easy 3. Romanian girls are not prostitutes 4. Romanian girls are not rude 5. Romanian girls actually do like Americans 1. Romanian girls are not gold diggers Many people think that Romanian girls are only interested in American guys because of the “gold” we have. The truth is, it’s not the money that Romanian girls are attracted to, it’s the person. 2.

:rofl:

Wow, am gasit ceva frumos. Un tool care genereaza portrete folosind inputs cum ar fi: statui, poze vechi, etc. (cere si asta cont, e gratuit aparent cu google).

Am bagat imaginea asta cu Caesar:
220px-Gaius_Iulius_Caesar_(Vatican_Museum)

Si a generat:
portrait-preview (2)

Haha, ce tare. Iacata-l pe Caesar!

6 Likes

Merge si invers ? Incearca pe Citu :slight_smile:

nu cred ca merge invers :slight_smile:

Socrate. :joy:

Seamana un pic cu Boris Johnson.
portrait-preview (3)

[Edit]

Burebista.

Rough patch saracul.
portrait-preview (4)

Vlad Tepes

portrait-preview (5)

Avram Iancu

avram_iancu

Ciprian Porumbescu

ciprian_porumbesu

Eminescu (idolul femeilor)

eminescu

Mihail Kogalniceanu

mihail_kogalniceanu

Nicolae Balcescu (figura renascentista)

nicolae_baslcescu

Stefan Cel Mare (rocker)

stefan_cel_mare

Vasile Ursu Nicola

vasile_ursu_nicola

(am scris si post pe blog)

3 Likes

Asta folosesc eu cand am nevoie de portrete:
https://thispersondoesnotexist.com/

https://huyenchip.com/ml-interviews-book/

2 Likes

Pare faina cartea. Huyen Chip tot mai mult continut super misto scoate.

1 Like
1 Like

http://nvidia-research-mingyuliu.com/gaugan

Daca tot e thread de discutie:

  • Cine de aici mai lucreaza cu ML pe chestii serioase/valoroase (adica nu doar pe un Proof of Concept ca o auzit managerul ca AI e la moda)?
  • Antrenati modele de la 0? Folositi ceva pretrained? Sau doar API-uri din cloud?
  • Pe ce domenii, computer vision, NLP, altceva?
  • Tensorflow sau Pytorch? :laughing:
1 Like

Salut,

Eu sunt curios daca exista cineva pe aici care este ML Engineer/ Data Scientist si ce presupune exact job-ul de zi cu zi(ce task-uri are, ce trebuie sa faca), similar cu intrebarile lui @rolisz de mai sus. Intreb asta, deoarece eu in prezent lucrez ca Java BE DEV, insa la facultate atat licenta, cat si disertatia am facut-o pe ML si mi-a placut destul de mult ce am facut si ma gandesc la idea de a face o schimbare in cariera.

1 Like