[spam][ot][spam][ignore] Enlrge Ur Pen1s
goal: figuring out how to write manual generation loop for speech2text model
one time offer you are the winner!!!!! one million dollar prize this time only !!!! free tickets once-in-a-lifetime perfect getaway click here to take our quick survey guaranteed win!!!
great rates on car insurance at https://github.com/huggingface/transformers/blob/main/src/transformers/model... special discount if you can identify the subparts of the main causal generation model! use the code EARLY_APRIL at checkout.
goal is figuring out token type propagation through generation loop worrying about other things. big worries.
found some parts near bottom of file, which does not provide car insurance, sorry if you needed it. best car insurance is to prioritise safety over all other things. huge returns. it has lm_head which is used for output tokens. can also be resized. but it transforms from d_model. should see where it is used in forward passes.
goal is something to do with generate(). looking to see what token inputs to provide, how to loop.
goal is to retain goal to completion. this goal has not previously been met, and is a very important skill, so continuing to attempt is valued.
big inhibition experience. the non-goal is to play a video game, or eat something yummy, or sleep. if inhibition-splash happens, can splash on one of these?
we have made a private area here for supporting goal in public the area is defended by the appearance of spam, when in reality it contains small personal notes to help me focus on what i'm doing in the moment. goal: stabilise goal of copying generate() parts out
[spam][ignore][spam][boring][tax return][bills] this goal is authorised and good: goal: copy generate() parts out
I will be collecting my spam replies for evolutionary entropy so as to produce more creative results. I am happy to learn that I am getting closer to better spam.
Need something exciting when it just won't get up? VIAGRA Make every morning a new day. goal: figure out how to move forward on speech-to-text generation. am passing wrong model dimension in. maybe just troubleshoot it.
What is good standalone (no cloud etc) opensource Linux app for speech2text?
On Sun, Apr 3, 2022, 2:09 AM grarpamp <grarpamp@gmail.com> wrote:
What is good standalone (no cloud etc) opensource Linux app for speech2text?
I don't know, but the gist I posted will run on an offline cpu (if the model is downloaded in advance by running the code while online, and then download disabled via setting TRANSFORMERS_OFFLINE in the environment), my devices were just all breaking while I coded it, so I stayed on colab.
I tried to add a little of the training loop parts at https://github.com/xloem/s2t.git . It's untested.
participants (5)
-
grarpamp
-
Karl Semich
-
Punk-BatSoup-Stasi 2.0
-
Undiscussed Horrific Abuse, One Victim of Many
-
zeynepaydogan