Checkpoint 2000
From Dontopedia, the open, paraconsistent wiki. (Last updated 2026-06-06.)
Checkpoint 2000 has 12 facts recorded in Dontopedia across 4 references, with 1 live disagreement.
Mostly:rdf:type(2), checkpointed at step(1), saved as(1)
Maturity scale
raw canonical shape-checked rule-derived certifiedOther facts (12)
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 |
|---|---|---|
| Rdf:type | Checkpoint | [2] |
| Rdf:type | Model Checkpoint | [4] |
| Checkpointed at Step | 2000 | [1] |
| Saved As | Step 002000 | [1] |
| Corresponds to Step | Training Step 2000 | [2] |
| Has Filename | step_002000 | [2] |
| File Path | /tmp/wave_native_ckpts_25m/step_2000/model.npz | [3] |
| File Format | npz | [3] |
| Scheduled at Step | 2000 | [4] |
| Estimated Time | ~14 minutes | [4] |
| Allows Inference | true | [4] |
| Checkpoint Path | /tmp/wave_native_ckpts_50m/step_2000/model.npz | [4] |
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 (4)
ctx:discord/blah/watt-activation/145- full textwatt-activation-145text/plain3 KB
doc:agent/watt-activation-145/64378aae-25d8-4d98-9efb-8655275675ddShow excerpt
[2026-03-09 15:14] xenonfun: recommending https://huggingface.co/datasets/HuggingFaceTB/everyday-conversations-llama3.1-2k for conversations at this gpt-2 small scale [2026-03-09 15:16] xenonfun: Practical recommendation: Start with everyda…
ctx:discord/blah/watt-activation/150- full textwatt-activation-150text/plain2 KB
doc:agent/watt-activation-150/337a9b46-3368-40e6-b1b8-d27104be216dShow excerpt
[2026-03-09 15:48] xenonfun: step 2000/16684 12.0% loss=6.0245 ppl= 413.4 lr=4.88e-05 667ms 12,290tok/s eta=163min VAL step 2000 loss=6.0279 ppl=414.9 ★ best ckpt step 2000 → step_002000 step 2100/16684 12.6% loss=6.01…
ctx:discord/blah/watt-activation/642- full textwatt-activation-642text/plain3 KB
doc:agent/watt-activation-642/3d74a9d0-e733-41fb-ac2c-53a6f761291fShow excerpt
[2026-04-16 02:04] xenonfun: ``` BPB 8.37 → 2.35 in 1,800 steps. Clean convergence, no instability. The first checkpoint at step 2,000 is about to save. Throughput steady at ~7.2K tok/s at batch=16 with Option B + RotationalAdamW. For…
ctx:discord/blah/watt-activation/646- full textwatt-activation-646text/plain2 KB
doc:agent/watt-activation-646/84b51c92-144f-432e-8454-c9f4c46ef821Show excerpt
[2026-04-16 03:27] xenonfun: ``` ⏺ Step 1,200: BPB 2.90 best, ~1,899 tok/s. Training well — BPB 8.6 → 2.9 in 1.2K steps. Throughput dipped slightly at step 1000 due to the DoReMi reference snapshot (cloning 50M params to disk + loading)…
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.