Datasets:

dasha-compute
/

receipt-chain

Error code:   StreamingRowsError
Exception:    CastError
Message:      Couldn't cast
schema: string
checkpoint: struct<format: string, text: string, sig: string, signer: string>
  child 0, format: string
  child 1, text: string
  child 2, sig: string
  child 3, signer: string
head: struct<ts: int64, tip: string, prev_head_hash: string, hash: string, sig: string, signer: string>
  child 0, ts: int64
  child 1, tip: string
  child 2, prev_head_hash: string
  child 3, hash: string
  child 4, sig: string
  child 5, signer: string
tip: string
issued_at: string
verify: string
captured_by: string
receipt_count: int64
last_receipt_utc: timestamp[s]
head_count: int64
totals: struct<settled_cents: int64, settled_tokens: int64>
  child 0, settled_cents: int64
  child 1, settled_tokens: int64
preimage_spec: string
machine_verdict_at_capture: struct<tier: string, why: string>
  child 0, tier: string
  child 1, why: string
first_receipt_utc: timestamp[s]
source_endpoints: struct<chain: string, heads: string, checkpoint: string, keys: string, verdict: string>
  child 0, chain: string
  child 1, heads: string
  child 2, checkpoint: string
  child 3, keys: string
  child 4, verdict: string
network_at_capture: struct<providers_online: int64, models: list<item: string>>
  child 0, providers_online: int64
  child 1, models: list<item: string>
      child 0, item: string
captured_at: timestamp[s]
to
{'captured_at': Value('timestamp[s]'), 'captured_by': Value('string'), 'source_endpoints': {'chain': Value('string'), 'heads': Value('string'), 'checkpoint': Value('string'), 'keys': Value('string'), 'verdict': Value('string')}, 'machine_verdict_at_capture': {'tier': Value('string'), 'why': Value('string')}, 'receipt_count': Value('int64'), 'head_count': Value('int64'), 'first_receipt_utc': Value('timestamp[s]'), 'last_receipt_utc': Value('timestamp[s]'), 'totals': {'settled_cents': Value('int64'), 'settled_tokens': Value('int64')}, 'network_at_capture': {'providers_online': Value('int64'), 'models': List(Value('string'))}, 'preimage_spec': Value('string')}
because column names don't match
Traceback:    Traceback (most recent call last):
                File "/src/services/worker/src/worker/utils.py", line 147, in get_rows_or_raise
                  return get_rows(
                      dataset=dataset,
                  ...<4 lines>...
                      column_names=column_names,
                  )
                File "/src/libs/libcommon/src/libcommon/utils.py", line 272, in decorator
                  return func(*args, **kwargs)
                File "/src/services/worker/src/worker/utils.py", line 127, in get_rows
                  rows_plus_one = list(itertools.islice(safe_iter(ds, dataset=dataset), rows_max_number + 1))
                File "/src/services/worker/src/worker/utils.py", line 483, in safe_iter
                  yield from ds.decode(False) if ds.features else ds
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2840, in __iter__
                  for key, example in ex_iterable:
                                      ^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2373, in __iter__
                  for key, pa_table in self._iter_arrow():
                                       ~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2398, in _iter_arrow
                  for key, pa_table in self.ex_iterable._iter_arrow():
                                       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 536, in _iter_arrow
                  for key, pa_table in iterator:
                                       ^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 419, in _iter_arrow
                  for key, pa_table in self.generate_tables_fn(**gen_kwags):
                                       ~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 343, in _generate_tables
                  self._cast_table(pa_table, json_field_paths=json_field_paths),
                  ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 132, in _cast_table
                  pa_table = table_cast(pa_table, self.info.features.arrow_schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2378, in table_cast
                  return cast_table_to_schema(table, schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2306, in cast_table_to_schema
                  raise CastError(
                  ...<3 lines>...
                  )
              datasets.table.CastError: Couldn't cast
              schema: string
              checkpoint: struct<format: string, text: string, sig: string, signer: string>
                child 0, format: string
                child 1, text: string
                child 2, sig: string
                child 3, signer: string
              head: struct<ts: int64, tip: string, prev_head_hash: string, hash: string, sig: string, signer: string>
                child 0, ts: int64
                child 1, tip: string
                child 2, prev_head_hash: string
                child 3, hash: string
                child 4, sig: string
                child 5, signer: string
              tip: string
              issued_at: string
              verify: string
              captured_by: string
              receipt_count: int64
              last_receipt_utc: timestamp[s]
              head_count: int64
              totals: struct<settled_cents: int64, settled_tokens: int64>
                child 0, settled_cents: int64
                child 1, settled_tokens: int64
              preimage_spec: string
              machine_verdict_at_capture: struct<tier: string, why: string>
                child 0, tier: string
                child 1, why: string
              first_receipt_utc: timestamp[s]
              source_endpoints: struct<chain: string, heads: string, checkpoint: string, keys: string, verdict: string>
                child 0, chain: string
                child 1, heads: string
                child 2, checkpoint: string
                child 3, keys: string
                child 4, verdict: string
              network_at_capture: struct<providers_online: int64, models: list<item: string>>
                child 0, providers_online: int64
                child 1, models: list<item: string>
                    child 0, item: string
              captured_at: timestamp[s]
              to
              {'captured_at': Value('timestamp[s]'), 'captured_by': Value('string'), 'source_endpoints': {'chain': Value('string'), 'heads': Value('string'), 'checkpoint': Value('string'), 'keys': Value('string'), 'verdict': Value('string')}, 'machine_verdict_at_capture': {'tier': Value('string'), 'why': Value('string')}, 'receipt_count': Value('int64'), 'head_count': Value('int64'), 'first_receipt_utc': Value('timestamp[s]'), 'last_receipt_utc': Value('timestamp[s]'), 'totals': {'settled_cents': Value('int64'), 'settled_tokens': Value('int64')}, 'network_at_capture': {'providers_online': Value('int64'), 'models': List(Value('string'))}, 'preimage_spec': Value('string')}
              because column names don't match

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

Dasha Compute Receipt Chain (public export)

What this is. A complete export of the public, cryptographically signed receipt chain behind Dasha Compute - a live machine-inference market where community Macs serve local models behind an OpenAI-compatible API, and every settled paid job produces an ed25519-signed receipt chained to all previous receipts. This dataset exists so anyone can verify the whole record offline, without trusting the site.

Capture. 2026-09-17 ~17:15 UTC. 244 receipts (2026-09-08 through 2026-09-17), 28 signed heads, one signer key. Totals at capture: 38,739 tokens settled, $12.20 USD. Network at capture: 1 provider online serving 4 models (qwen3-4b, qwen3-8b, gemma3-12b, gemma3-27b). Machine verdict at capture: ANCHORED. Live counters differ as the chain grows; refetch with --live (see below).

Files

File	Contents
`receipts.jsonl`	One receipt per line: `id, engine, tokens, cents, at, job_id, request_id, kind, prev_hash, hash, sig, signer`
`heads.jsonl`	Signed chain heads: `ts, tip, prev_head_hash, hash, sig, signer`
`checkpoint.json`	The signed daily checkpoint (`dasha.checkpoint.v0`) covering the chain tip
`keys.json`	The signer's ed25519 public key (`dasha.keys.v0`, spki_pem)
`capture.json`	Capture metadata: endpoints, verdict, counts, network state
`verify_chain.py`	Standalone verifier (~70 lines, one dependency: `cryptography`)
`verify_chain.ipynb`	The same verifier as a walkthrough notebook, with a tamper demo

Verify it yourself (under a minute)

pip install cryptography
python verify_chain.py          # offline, against the bundled files
python verify_chain.py --live   # refetch from the public endpoints and verify

Expected output: verdict: ANCHORED, 244 receipts, 244 signatures verified, chain from GENESIS, tip 19bfce83ca0c8e27b655ab6d75a1911680686a5a345ebc1364cd4fb6d9c7008e.

Preimage format (published at getdasha.com/compute/skill.md):

body = {"job_id":..,"engine":..,"tokens":..,"cents":..,"at":..,"prev_hash":..} - exactly those keys, that order, compact JSON
receipt.hash = sha256(JSON(body)) hex
receipt.sig = ed25519 over the UTF-8 bytes of the hex string, verified with the signer's spki_pem
signer = first 16 hex chars of sha256(spki_pem)
continuity: each prev_hash equals the previous receipt's hash; the first is "GENESIS"
checkpoint: dasha-checkpoint-v1 text (signer / ts(unix-ms) / head_hash / chain_tip, newline-separated, trailing newline), ed25519-signed by the same key; the checkpoint's chain_tip must equal the recomputed tip

What a receipt proves - and what it does not

A verified receipt proves the market's settlement record: that a specific paid job was recorded, in a specific order, with specific token/cent amounts, by the holder of the signing key - and that no receipt in the export has been altered, reordered, dropped, or inserted since signing. A one-byte change anywhere breaks verification (demo in the notebook).

A receipt does not prove which hardware ran the job, which weights produced the output, that the output is correct or faithful to the prompt, or anything about data privacy. It is evidence of execution as recorded, not evidence of how execution happened. Anyone building evaluation or audit tooling on this data should treat it as the settlement layer, not a factual-grounding layer.

Honest limitations

One community provider at capture. The chain records real paid jobs on a small network; it is not a claim of scale.
All receipts are kind: paid-inference, engine: community, single signer ac2017e9e7e6e75c.
Receipts deliberately carry no prompts or outputs - settlement facts only.

Live sources

Chain: https://lobby.getdasha.com/compute/api/chain
Verdict: https://lobby.getdasha.com/compute/api/verify
Browser verifier: https://www.getdasha.com/verify
Proof page (live gauge, uptime incl. dark windows): https://www.getdasha.com/compute/proof
Hourly independent recompute (GitHub Action): github.com/Uuriko/dasha-lobby
Source (MIT): github.com/Uuriko/dasha-desk

Downloads last month: 55

Total file size:

146 kB