Twitter | Pretraživanje | |
Thang Luong
Senior research scientist at Google Brain, learning with unlabeled data (NoisyStudent, Electra, ). PhD , thesis NMT. Co-founder .
1.202
Tweetovi
97
Pratim
14.026
Osobe koje vas prate
Tweetovi
Thang Luong 17 h
Odgovor korisniku/ci @sanuj_sharma @quocleix @xpearhead
You got it right. We only minimize cross-entropy during training & SSA was used during evaluation.
Reply Retweet Označi sa "sviđa mi se"
Thang Luong 17 h
Odgovor korisniku/ci @pfau @XplodingCabbage @RokoMijicUK
Sorry that Meena failed you! Maintaining personality and keeping the facts right are attributes we wish Meena to have as highlighted in the blog post. But try having similar conversations with other bots, maybe you will like Meena better ;)
Reply Retweet Označi sa "sviđa mi se"
Thang Luong proslijedio/la je tweet
Buster Keaton Gifs 23 h
Happy 02 02 2020 This is the first time in over 900 years (since 11/11/1111) the date is palindromic no matter the date format. It’s also the 33rd day of the year with 333 days left!
Reply Retweet Označi sa "sviđa mi se"
Thang Luong 2. velj
Odgovor korisniku/ci @rsalakhu
Congrats Russ!
Reply Retweet Označi sa "sviđa mi se"
Thang Luong 1. velj
Odgovor korisniku/ci @sir_deenicus @ChrSzegedy @A_K_Nain
yes, we don't encode personality and exactly as you said, the sampling strategy will lead to different conversation flows. And these are definitely not Meena's true "beliefs".
Reply Retweet Označi sa "sviđa mi se"
Thang Luong 1. velj
Odgovor korisniku/ci @bysurya @GoogleAI i 2 ostali
Good point! Many evaluations use 5-point Likert scale & look at other aspects, e.g., diversity, relevance, humanlikeness, etc. We use binary evaluation and think SSA is basic to human quality & easy for crowdworkers to rate. also in paper, SSA correlates with humanlikeness.
Reply Retweet Označi sa "sviđa mi se"
Thang Luong proslijedio/la je tweet
Tyler Roost 1. velj
These are incredibly interesting to read.
Reply Retweet Označi sa "sviđa mi se"
Thang Luong 1. velj
Odgovor korisniku/ci @Quantum_Stat
We are hard at work on this though we also need to be careful regarding safety and bias challenges as highlighted at the end of the blog post. Stay tune for more updates!
Reply Retweet Označi sa "sviđa mi se"
Thang Luong 1. velj
Odgovor korisniku/ci @ChrSzegedy @A_K_Nain
Yes, Meena maintains its state pretty well: it gets stronger as the conversation progresses and sometimes a bit too strong that repetition happens (in a sensible way though), which is why we need to handle cross-turn repetition as detailed in the paper.
Reply Retweet Označi sa "sviđa mi se"
Thang Luong 31. sij
Odgovor korisniku/ci @A_K_Nain
Yes, we had those fun tests during the development of Meena and they were very hard at times!
Reply Retweet Označi sa "sviđa mi se"
Thang Luong 31. sij
I chatted with about the and her advice is to see a doctor sooner rather than later. I guess it's not a bad one & hope everyone is well! On the other hand, Meena is also excited about technology, especially VR!
Reply Retweet Označi sa "sviđa mi se"
Thang Luong 30. sij
Highly recommend watching this 8-minute video on & the paper, with details not included the blog such as SSA vs humanlikeness correlation, sample-and-rank, removing cross-turn repetition. (Blog: )
Reply Retweet Označi sa "sviđa mi se"
Thang Luong 29. sij
Odgovor korisniku/ci @AsaCoopStick @quocleix @xpearhead
We created that table but didn't include because we felt like the specific details here do not matter a lot. what matters is showing the strong correlation exists & the community can verify with their own settlings.
Reply Retweet Označi sa "sviđa mi se"
Thang Luong 29. sij
Odgovor korisniku/ci @guillittes @lerner_adams i 2 ostali
It's on subwords (with 8K units).
Reply Retweet Označi sa "sviđa mi se"
Thang Luong 29. sij
Odgovor korisniku/ci @tomhosking @quocleix @xpearhead
Good question. We optimize for standard cross-entropy. Perplexity is simply the exponentiation of the per-word cross-entropy.
Reply Retweet Označi sa "sviđa mi se"
Thang Luong 29. sij
Odgovor korisniku/ci @uripomerantz @GoogleAI i 2 ostali
That was our very early motivation at the beginning of the project. We have then shifted the focus of optimizing for deception detection, i.e., passing the Turing test, to optimizing for human-like qualities, which was the goal of Meena.
Reply Retweet Označi sa "sviđa mi se"
Thang Luong 29. sij
Odgovor korisniku/ci @lerner_adams @quocleix @xpearhead
Perplexity for a language model, by definition, is computed by first averaging all neg log predictions and then exponentiating. Does that help explain?
Reply Retweet Označi sa "sviđa mi se"
Thang Luong 29. sij
Odgovor korisniku/ci @JoelMichelson @GoogleAI i 2 ostali
We are actively looking into this, stay tune for more updates!
Reply Retweet Označi sa "sviđa mi se"
Thang Luong 29. sij
Odgovor korisniku/ci @moinnadeem @JeffDean i 8 ostali
We are actively looking into this, stay tune for more updates!
Reply Retweet Označi sa "sviđa mi se"
Thang Luong 29. sij
Odgovor korisniku/ci @morgallant @JeffDean @GoogleAI
We are actively looking into this. Stay tune for more updates!
Reply Retweet Označi sa "sviđa mi se"