OverviewFilesCommitsIssuesDocsReleasesPeers

opencog-learn commits / a0ca64d

Resume planning prep
a0ca64d5f2ff1bcaff1de6e22842f44c9a644f8e
by Linas Vepštas <linasvepstas@gmail.com> · 2026-01-15T21:32:23.000Z

Changed files (1) +129 -0

learn-lang-diary/learn-lang-diary-part-ten-F.lyx +129 -0
……@@ -3341,6 +3341,135 @@ Anyway,
33413341 and discarded when the session ends.
33423342\end_layout
33433343
3344\begin_layout Subsection*
334515 January 2026
3346\end_layout
3347
3348\begin_layout Standard
3349Well,
3350 i got interrupted while writing the above,
3351 and once again I am not interested in finishing.
3352 Except as a planning exercise.
3353 Let me list everything on the plan.
3354\end_layout
3355
3356\begin_layout Itemize
3357Work on what it means to
3358\begin_inset Quotes eld
3359\end_inset
3360
3361remember something
3362\begin_inset Quotes erd
3363\end_inset
3364
3365,
3366 including the reduction of
3367\begin_inset Quotes eld
3368\end_inset
3369
3370all related things
3371\begin_inset Quotes erd
3372\end_inset
3373
3374 to a shorter list of
3375\begin_inset Quotes eld
3376\end_inset
3377
3378relevant things
3379\begin_inset Quotes erd
3380\end_inset
3381
3382,
3383 to
3384\begin_inset Quotes eld
3385\end_inset
3386
3387focus attention on those relevant things.
3388 This is actually a fascinating project,
3389 and the above description from two days ago,
3390 was about a project I started to do exactly this,
3391 with Claude.
3392 It is interesting,
3393 its useful,
3394 its important.
3395 So why not do it?
3396 I fear a dead end.
3397 I was asking Claude to build this for me,
3398 and it was an endless stream of disappointments.
3399 Claude is stunningly stupid in some very serious ways.
3400 Plus,
3401 as a proprietary system,
3402 interfacing with it,
3403 the cost,
3404 the challenges of a context window that I have poor or no management over,
3405 these are problems.
3406 I could replace Claude by some open LLM I could run locally...
3407 use Claude for design,
3408 but integrate into the local system...
3409\end_layout
3410
3411\begin_layout Itemize
3412I can resume the word–pair counting project.
3413 Now that I have built up enough infrastructure,
3414 it should be resumable.
3415 But it also has issues,
3416 see below.
3417\end_layout
3418
3419\begin_layout Itemize
3420Jizz up some interfaces for audio and photographs.
3421 So that when I resume word–counting,
3422 I would also deploy on audio and video.
3423\end_layout
3424
3425\begin_layout Itemize
3426Continue work on SIMD,
3427 which is about the abstraction and manipulation of structures
3428\begin_inset Quotes eld
3429\end_inset
3430
3431external to self
3432\begin_inset Quotes erd
3433\end_inset
3434
3435:
3436 structures that have internal descriptive forms,
3437 in Atomese,
3438 but require performance,
3439 control and execution in a
3440\begin_inset Quotes eld
3441\end_inset
3442
3443remote place
3444\begin_inset Quotes erd
3445\end_inset
3446
3447,
3448 the GPU for this particular situation.
3449\end_layout
3450
3451\begin_layout Standard
3452So the question becomes:
3453 what should I work on first?
3454 The answer is,
3455 perhaps,
3456 all of them,
3457 all at once,
3458 in parallel.
3459 I was gong to write much more here,
3460 explaining each,
3461 reviewing each,
3462 weighing the pros and cons of each,
3463 but then I interrupted myself to chit–chat on discord,
3464 and now I'm chit–chatted out and don't want to write any more,
3465 and instead just start doing,
3466 instead.
3467 So,
3468 fuck it.
3469 I'm done here.
3470 The chat log is below.
3471\end_layout
3472
33443473\begin_layout Section*
33453474The End
33463475\end_layout