context

Part 25

ctx:discord/blah/random/part-25

kind: custommode: permissive104 statements57 distinct subjects

No external document is attached to this context. (Many contexts are pure organisational labels.)

Facts in this context

Grouped by subject. Each subject links to its full article.

Ajaxdavis17 facts`ex:ajaxdavis`

evaluatesNovelsAs	Pretty Damn Good
expressesLaughter	lmao
hasRead	Quite a Bit
isTrainingJsModelOn	Novels Dataset
postedMessageAt	2026-02-17 09:26
postedMessageAt	2026-02-17 11:21
postedMessageAt	2026-02-17 10:26
postedMessageAt	2026-02-17 10:10
postedMessageAt	2026-02-17 10:07
postedMessageAt	2026-02-17 09:54
postedMessageAt	2026-02-17 07:21
prefersNovelsOverShakespeareData	True
quotesParaphraseOf	to be or not to bed and he not
saysWord	openclaw
sharesBetterNovels	Novels Dataset
startedVersionNamed	alpha
wroteScriptWith	Claude

Xenonfun10 facts`ex:xenonfun`

advisesTo	Hit Patience 10 or Ctrl C
believesGradientAccumulationIsOptimal	True
postedMessageAt	2026-02-17 05:10
postedMessageAt	2026-02-17 09:21
postedMessageAt	2026-02-17 10:04
postedMessageAt	2026-02-17 11:07
postedReplyAt	2026-02-17 09:28
recommendsCommand	Randygpt Command
recommendsNextResumeWith	Higher Iter Count
recommendsResumeWith	Bigger Iter Target

Bpe Process4 facts`ex:bpe-process`

isMultithreaded	True
runsOn	Cpu
savesVocab	True
takesTime	15

Gradient Accumulation4 facts`ex:gradient-accumulation`

addressesExactly	This Problem
enablesHigherBaseLr	Without Instability
isRightFixFor	Current Training Issue
producesMoreStableGradients	Gradients

Model4 facts`ex:model`

hasLSizeMB	66
hasMParams	31
trainsAtFullLrFromIter	1000
trainsAtFullLrToIter	1800

1k Run3 facts`ex:1k-run`

estimatedCompletionIn	30
expectedToBeMuchBetterThan	Best Run Before
forApplesToApplesCompare	Previous Run

Novels Dataset3 facts`ex:novels-dataset`

containsNumNovels	38
locatedAtPath	/home/ajax/repos/tpmjs/discord-books/novels/
totalWords	280509

Traves Theberge3 facts`ex:traves-theberge`

declaresZoneAs	Claw Free Zone
mentionsUserId	741226685969596429
postedMessageAt	2026-02-17 06:59

Best Checkpoint2 facts`ex:best-checkpoint`

atIteration	728
named	checkpoint_best.bin

Current Memory Usage2 facts`ex:current-memory-usage`

atLimitOf	Gpu Hardware
spikesToApproximately	12

Decay Start Iter2 facts`ex:decay-start-iter`

atIter	1800
equals60PercentOf	3000

Effective Batch Size2 facts`ex:effective-batch-size`

equals	256
inTokensPerStep	256

Lisamegawatts2 facts`ex:lisamegawatts`

evaluatesShakespeareAs	Great Writer
postedMessageAt	2026-02-17 11:16

Xenonfun 10 04 Message2 facts`ex:xenonfun-10-04-message`

includesAttachments	Screenshot 2026 02 17 at 5 04 16 Am Png
includesAttachments	Screenshot 2026 02 17 at 5 01 50 Am Png

Xenonfun Shakespeare Dataset2 facts`ex:xenonfun-shakespeare-dataset`

containsAllOf	Shakespeare
sizeApproximately	7

Ajaxdavis 09 26 Message1 fact`ex:ajaxdavis-09-26-message`

asksAbout

Gpu Mem

Ajaxdavis 10 10 Message1 fact`ex:ajaxdavis-10-10-message`

asksQuestionAbout

Xenonfun Training Data Word Count

Ajaxdavis 10 26 Message1 fact`ex:ajaxdavis-10-26-message`

includesAttachment

Books Zip

Ajaxdavis 11 21 Message1 fact`ex:ajaxdavis-11-21-message`

includesAttachment

Novels Zip

Ajaxdavis Script1 fact`ex:ajaxdavis-script`

convertsEveryChannelInServerTo

Book

Alpha Version1 fact`ex:alpha-version`

implementedIn

typescript

Bpe Early Results1 fact`ex:bpe-early-results`

evaluatedAsMuchBetter

Previous Results

Bpe Tokenization1 fact`ex:bpe-tokenization`

improvesPerformanceOver

Char Tokening

Char Tokening1 fact`ex:char-tokening`

causedStallOut

Training

Chat1 fact`ex:chat`

referencesShakespeare

Multiple Times

Chat Participants1 fact`ex:chat-participants`

shareModelTrainingProgress

True

Claude1 fact`ex:claude`

assistedInWriting

Ajaxdavis Script

Claw Joke1 fact`ex:claw-joke`

madeByTraves

True

Cpu Side Memory1 fact`ex:cpu-side-memory`

needsAbout

600

Current Run1 fact`ex:current-run`

isCoastingDownhillOn

Spent Lr Budget

Current Training Issue1 fact`ex:current-training-issue`

presupposedBy

Xenonfun First Message

Discord Books1 fact`ex:discord-books`

totalWordsApproximately

750000

Discord Server1 fact`ex:discord-server`

channelsConvertedToBooksBy

Ajaxdavis Script

Gpu1 fact`ex:gpu`

isHardwareLimitForTraining

True

Gpu Hardware1 fact`ex:gpu-hardware`

hasMaxMemory

Grad Accum Steps1 fact`ex:grad-accum-steps`

setToValue

Js Model1 fact`ex:js-model`

trainedBy

Ajaxdavis

Learning1 fact`ex:learning`

happensInRange

1000 to 1800 Iters

Lr Decay1 fact`ex:lr-decay`

startsAt60PercentOfTotalIters

True

Lr Schedule1 fact`ex:lr-schedule`

givesMoreRoomToModel

With Higher Iter Target

Model Training1 fact`ex:model-training`

usesLrDecaySchedule

True

Openclaw1 fact`ex:openclaw`

respondsToClawJoke

True

Optimizer Steps1 fact`ex:optimizer-steps`

frequencyReducedByFactor

Patience1 fact`ex:patience`

setTo

Previous Peak Memory1 fact`ex:previous-peak-memory`

approximately

Randygpt1 fact`ex:randygpt`

supportsResumeFlag

True

Randygpt Command1 fact`ex:randygpt-command`

specifiedAs

./target/release/randygpt --iters 3000 --resume

Resume Training to 2000 Iters1 fact`ex:resume-training-to-2000-iters`

providesMoreRoomFor

Decay Schedule

Shakespeare1 fact`ex:shakespeare`

usedAsTrainingData

True

Small Model Checkpoint1 fact`ex:small-model-checkpoint`

sizeInMB

9.7

Tokenization1 fact`ex:tokenization`

movedFromCharTo

Bpe

Training1 fact`ex:training`

involvesResumingCheckpoints

True

Training With Bpe1 fact`ex:training-with-bpe`

willSucceedBetter

True

Weights at Iter 7281 fact`ex:weights-at-iter-728`

areBestState

Checkpoint Best Bin

Xenonfun 09 21 Message1 fact`ex:xenonfun-09-21-message`

includesAttachment

Screenshot 2026 02 17 at 4 20 17 Am Png

Xenonfun 09 28 Message1 fact`ex:xenonfun-09-28-message`

isReplyTo

Ajaxdavis 09 26 Message

Zone1 fact`ex:zone`

permitsOnly

Dogs

Part 25

Facts in this context

Ajaxdavis17 factsex:ajaxdavis

Xenonfun10 factsex:xenonfun

Bpe Process4 factsex:bpe-process

Gradient Accumulation4 factsex:gradient-accumulation

Model4 factsex:model

1k Run3 factsex:1k-run

Novels Dataset3 factsex:novels-dataset

Traves Theberge3 factsex:traves-theberge

Best Checkpoint2 factsex:best-checkpoint

Current Memory Usage2 factsex:current-memory-usage

Decay Start Iter2 factsex:decay-start-iter

Effective Batch Size2 factsex:effective-batch-size

Lisamegawatts2 factsex:lisamegawatts

Xenonfun 10 04 Message2 factsex:xenonfun-10-04-message

Xenonfun Shakespeare Dataset2 factsex:xenonfun-shakespeare-dataset

Ajaxdavis 09 26 Message1 factex:ajaxdavis-09-26-message

Ajaxdavis 10 10 Message1 factex:ajaxdavis-10-10-message

Ajaxdavis 10 26 Message1 factex:ajaxdavis-10-26-message

Ajaxdavis 11 21 Message1 factex:ajaxdavis-11-21-message

Ajaxdavis Script1 factex:ajaxdavis-script

Alpha Version1 factex:alpha-version

Bpe Early Results1 factex:bpe-early-results

Bpe Tokenization1 factex:bpe-tokenization

Char Tokening1 factex:char-tokening

Chat1 factex:chat

Chat Participants1 factex:chat-participants

Claude1 factex:claude

Claw Joke1 factex:claw-joke

Cpu Side Memory1 factex:cpu-side-memory

Current Run1 factex:current-run

Current Training Issue1 factex:current-training-issue

Discord Books1 factex:discord-books

Discord Server1 factex:discord-server

Gpu1 factex:gpu

Gpu Hardware1 factex:gpu-hardware

Grad Accum Steps1 factex:grad-accum-steps

Js Model1 factex:js-model

Learning1 factex:learning

Lr Decay1 factex:lr-decay

Lr Schedule1 factex:lr-schedule

Model Training1 factex:model-training

Openclaw1 factex:openclaw

Optimizer Steps1 factex:optimizer-steps

Patience1 factex:patience

Previous Peak Memory1 factex:previous-peak-memory

Randygpt1 factex:randygpt

Randygpt Command1 factex:randygpt-command

Resume Training to 2000 Iters1 factex:resume-training-to-2000-iters

Shakespeare1 factex:shakespeare

Small Model Checkpoint1 factex:small-model-checkpoint

Tokenization1 factex:tokenization

Training1 factex:training

Training With Bpe1 factex:training-with-bpe

Weights at Iter 7281 factex:weights-at-iter-728

Xenonfun 09 21 Message1 factex:xenonfun-09-21-message

Xenonfun 09 28 Message1 factex:xenonfun-09-28-message

Zone1 factex:zone

Ajaxdavis17 facts`ex:ajaxdavis`

Xenonfun10 facts`ex:xenonfun`

Bpe Process4 facts`ex:bpe-process`

Gradient Accumulation4 facts`ex:gradient-accumulation`

Model4 facts`ex:model`

1k Run3 facts`ex:1k-run`

Novels Dataset3 facts`ex:novels-dataset`

Traves Theberge3 facts`ex:traves-theberge`

Best Checkpoint2 facts`ex:best-checkpoint`

Current Memory Usage2 facts`ex:current-memory-usage`

Decay Start Iter2 facts`ex:decay-start-iter`

Effective Batch Size2 facts`ex:effective-batch-size`

Lisamegawatts2 facts`ex:lisamegawatts`

Xenonfun 10 04 Message2 facts`ex:xenonfun-10-04-message`

Xenonfun Shakespeare Dataset2 facts`ex:xenonfun-shakespeare-dataset`

Ajaxdavis 09 26 Message1 fact`ex:ajaxdavis-09-26-message`

Ajaxdavis 10 10 Message1 fact`ex:ajaxdavis-10-10-message`

Ajaxdavis 10 26 Message1 fact`ex:ajaxdavis-10-26-message`

Ajaxdavis 11 21 Message1 fact`ex:ajaxdavis-11-21-message`

Ajaxdavis Script1 fact`ex:ajaxdavis-script`

Alpha Version1 fact`ex:alpha-version`

Bpe Early Results1 fact`ex:bpe-early-results`

Bpe Tokenization1 fact`ex:bpe-tokenization`

Char Tokening1 fact`ex:char-tokening`

Chat1 fact`ex:chat`

Chat Participants1 fact`ex:chat-participants`

Claude1 fact`ex:claude`

Claw Joke1 fact`ex:claw-joke`

Cpu Side Memory1 fact`ex:cpu-side-memory`

Current Run1 fact`ex:current-run`

Current Training Issue1 fact`ex:current-training-issue`

Discord Books1 fact`ex:discord-books`

Discord Server1 fact`ex:discord-server`

Gpu1 fact`ex:gpu`

Gpu Hardware1 fact`ex:gpu-hardware`

Grad Accum Steps1 fact`ex:grad-accum-steps`

Js Model1 fact`ex:js-model`

Learning1 fact`ex:learning`

Lr Decay1 fact`ex:lr-decay`

Lr Schedule1 fact`ex:lr-schedule`

Model Training1 fact`ex:model-training`

Openclaw1 fact`ex:openclaw`

Optimizer Steps1 fact`ex:optimizer-steps`

Patience1 fact`ex:patience`

Previous Peak Memory1 fact`ex:previous-peak-memory`

Randygpt1 fact`ex:randygpt`

Randygpt Command1 fact`ex:randygpt-command`

Resume Training to 2000 Iters1 fact`ex:resume-training-to-2000-iters`

Shakespeare1 fact`ex:shakespeare`

Small Model Checkpoint1 fact`ex:small-model-checkpoint`

Tokenization1 fact`ex:tokenization`

Training1 fact`ex:training`

Training With Bpe1 fact`ex:training-with-bpe`

Weights at Iter 7281 fact`ex:weights-at-iter-728`

Xenonfun 09 21 Message1 fact`ex:xenonfun-09-21-message`

Xenonfun 09 28 Message1 fact`ex:xenonfun-09-28-message`

Zone1 fact`ex:zone`