Dataset Viewer
Auto-converted to Parquet Duplicate
checks_run
list
error_count
int64
errors
list
format_version
int64
receipt_sha256
string
schema
string
structural_status
string
subject
string
warning_count
int64
warnings
list
[ "pre-seal" ]
0
[]
1
8390cf1ddd67bff4cbc87eef4b2e734b242063e9fb63dbbadd38f8ac9b93b836
malaiwah.fidelity-structural-validation.v1
sealed
dataset:fidelity--k2-horizon.community-root
0
[]

k2-horizon random CPU fixture root

A root fidelity dataset in hidden form, produced by engines/tools/hf_capture.py from malaiwah/k2-horizon-tiny-random-bf16.

The cut

the final hidden state handed to lm_head -- after the text model's final norm and immediately before the head matmul -- captured as the head module's input via torch.nn.Module.register_forward_pre_hook; replay applies the head ONLY (no final norm at replay time: the capture already sits after it). Same cut as engines/tools/hidden_replay.py and as Festr's kimi-k3 hidden-replay qualification.

What is here

records 4
scored positions 252
context length 64
hidden width 64
vocab size 272
capture bytes (hidden form) 33416
the same capture in logit form would be 274176 bytes (8x)
scope policy native
lane other

Does the model still generate sensibly

The capital of France is -> (p=0.0056, entropy 5.593 of a possible 5.606 nats). Recorded, not enforced. One extra window through the same schedule; its hidden state was discarded and is not part of this dataset.

License

This dataset redistributes the captured model's native output-head weights. The source license and required notice are reproduced byte-for-byte in LICENSE.

How to check it

bin/fidelity-dataset verify <this directory>

Full specification: https://github.com/malaiwah/quant-fidelity-suite/blob/main/docs/FIDELITY-DATASET-SPEC.md

Downloads last month
26

Collections including malaiwah/k2-horizon-tiny-fidelity-root-v1