Current Training Run
From Dontopedia, the open, paraconsistent wiki. (Last updated 2026-05-02.)
Current Training Run has 34 facts recorded in Dontopedia across 5 references, with 2 live disagreements.
Mostly:total steps(3), at step(2), expected to finish in(1)
Maturity scale
raw canonical shape-checked rule-derived certifiedInbound mentions (19)
Other subjects in dontopedia point AT this entity as a value. These are inverse relationships — e.g. "X motherOf this subject" — and answer questions the forward facts can't. Grouped by predicate.
partOfPart of(3)
- All Steps
ex:all-steps - N Dim Sphere Attention
ex:n-dim-sphere-attention - Training Log 30500
ex:training-log-30500
achievedInTrainingAchieved in Training(1)
- Two Full Passes
ex:two-full-passes
at170wAt170w(1)
- Power Consumption
ex:power-consumption
conditionedTaskOnConditioned Task on(1)
- Xenonfun
ex:xenonfun
continuesTrainingContinues Training(1)
- Xenonfun
ex:xenonfun
decreasesOverItersDecreases Over Iters(1)
- Learning Rate
ex:learning-rate
equals0.29-dollarsEquals0.29 Dollars(1)
- Training Cost
ex:training-cost
fluctuatesAcrossStepsFluctuates Across Steps(1)
- Loss Metric
ex:loss-metric
for145m-model-two-epochsFor145m Model Two Epochs(1)
- 29 Cents Cost
ex:29-cents-cost
hostsImplementationHosts Implementation(1)
- Mlx
ex:mlx
introducesBugsIntroduces Bugs(1)
- AI Assistant
ex:ai-assistant
misinterpretsNonstandardMisinterprets Nonstandard(1)
- AI Assistant
ex:ai-assistant
preferredOverCurrentPreferred Over Current(1)
- Prior Training Run
ex:prior-training-run
previousVersionOfPrevious Version of(1)
- Model V7
ex:model-v7
reportsMetricsReports Metrics(1)
- Training Log
ex:training-log
Other facts (34)
The long tail: predicates that appear too rarely to warrant their own section. Filter or scroll to find a specific one. Each row links to its source.
| Predicate | Value | Ref |
|---|---|---|
| Total Steps | 16684 | [3] |
| Total Steps | 20000 | [4] |
| Total Steps | 1245 | [5] |
| At Step | 6600 | [4] |
| At Step | 25 | [5] |
| Expected to Finish in | 30m | [1] |
| Resumed55k Run Time | ~8h | [2] |
| Achieves Two Epochs | 145m Param Model | [2] |
| Uses Custom Vocab | Bigger Vocab | [2] |
| After Two Full Passes Provides | Rough Understanding of Language | [2] |
| First10k Run Time | ~1.5h | [2] |
| For Two Full Epochs | 145m Param Model | [2] |
| Trains145m Param Model | 145m Param Model | [2] |
| Total Run Time | ~9.5 hours | [2] |
| Resumed After First10k | 55k Continuation | [2] |
| Total Iters | 65000 | [2] |
| Uses Optimizer | Modified Adam | [3] |
| Deviates From | Standard Architecture | [3] |
| Implemented in | Mlx | [3] |
| Ongoing | true | [3] |
| Part of | Anchor Kan | [3] |
| Steadily Progressing | true | [3] |
| Uses Fifty Percent Less Flops | Anchor Kan 65k Run | [3] |
| Progress Percent | 2.0% | [5] |
| Has Loss | 4.4653 | [5] |
| Has Eta | 50min | [5] |
| Has DC | 0.06 | [5] |
| Has Lr | 5.00e-03 | [5] |
| Has R | 0.225 | [5] |
| Has Snr | -12.7dB | [5] |
| Has Temp | 0.76 | [5] |
| Has Throughput | 212,398 tok/s | [5] |
| Has C | 2.7b | [5] |
| Has Bpb | 6.442 | [5] |
Timeline
Timeline axis is valid_time — when each source says the fact was true in the world, not when Dontopedia learned about it. Retracted rows are kept for provenance; coloured stripes indicate the context kind.
References (5)
ctx:discord/blah/safiersemantics/part-72ctx:discord/blah/watt-activation/part-97ctx:discord/blah/watt-activation/part-153ctx:discord/blah/watt-activation/part-373ctx:discord/blah/watt-activation/part-406
See also
Keep researching
Missing something or suspicious of what's here? Kick off a research session — a Claude agent will investigate, cite its sources, and file new facts into a dedicated context you can review before accepting into the shared view.