nickh007 commited on
Commit
96854e9
·
verified ·
1 Parent(s): 8bd6377

Publish dataset with card, data and loader

Browse files
Files changed (2) hide show
  1. CITATION.cff +16 -0
  2. README.md +70 -0
CITATION.cff ADDED
@@ -0,0 +1,16 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ cff-version: 1.2.0
2
+ message: If you use this software, please cite it as below.
3
+ title: "abstain-corpus: 32 inputs a verifier must NOT pass"
4
+ abstract: >-
5
+ 32 inputs a verifier must NOT pass
6
+ authors:
7
+ - family-names: Harris
8
+ given-names: Nick
9
+ repository-code: "https://github.com/nickharris808/abstain-corpus"
10
+ url: "https://github.com/nickharris808/abstain-corpus"
11
+ license: Apache-2.0
12
+ type: software
13
+ keywords:
14
+ - verification
15
+ - abstention
16
+ - reproducibility
README.md CHANGED
@@ -115,3 +115,73 @@ scored cases, report an exact binomial interval rather than a bare proportion: 0
115
 
116
  Apache-2.0. Generated by [`abstain-bench`](https://github.com/nickharris808/abstain-bench)
117
  (`abstain_bench.corpus.build_corpus`).
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
115
 
116
  Apache-2.0. Generated by [`abstain-bench`](https://github.com/nickharris808/abstain-bench)
117
  (`abstain_bench.corpus.build_corpus`).
118
+
119
+ <!-- PORTFOLIO -->
120
+ ---
121
+
122
+ ## The rest of the portfolio
123
+
124
+ 25 artifacts, one idea: **a measurement you cannot check is a press release.** Every tool
125
+ here reports; none of them gates.
126
+
127
+ **Tools**
128
+
129
+ | | |
130
+ |---|---|
131
+ | [`abstain-bench`](https://github.com/nickharris808/abstain-bench) | how often does a verifier pass input it could not check? |
132
+ | [`evidence`](https://github.com/nickharris808/evidence) | run the whole portfolio over your repo — the weakest leg, never the mean |
133
+ | [`floorgen`](https://github.com/nickharris808/floorgen) | what must your system remember? an exact lower bound |
134
+ | [`formal-proof-mcp`](https://github.com/nickharris808/formal-proof-mcp) | a proof kernel for your coding agent |
135
+ | [`gatecount`](https://github.com/nickharris808/gatecount) | exactly how many states does removing this check admit? |
136
+ | [`gridlock`](https://github.com/nickharris808/gridlock) | certify a wait-for relation cannot wedge |
137
+ | [`honestbench`](https://github.com/nickharris808/honestbench) | measure your CI's escape rate |
138
+ | [`kvleak`](https://github.com/nickharris808/kvleak) | cross-tenant leak scanner |
139
+ | [`kvprobe`](https://github.com/nickharris808/kvprobe) | model-substitution detector with a measured FPR |
140
+ | [`preregister`](https://github.com/nickharris808/preregister) | refuses to seal a plan whose conclusion is already fixed |
141
+ | [`proof-carrying-ci`](https://github.com/nickharris808/proof-carrying-ci) | the whole portfolio as one CI check, with SARIF |
142
+ | [`proof-to-code-drift`](https://github.com/nickharris808/proof-to-code-drift) | fail the build when the proof stops matching |
143
+ | [`sf-verify`](https://github.com/nickharris808/sf-verify) | re-derive admission decisions offline |
144
+ | [`signoff-cert`](https://github.com/nickharris808/signoff-cert) | certificates that carry their own false-pass bound |
145
+ | [`tokencount`](https://github.com/nickharris808/tokencount) | a token count both parties can recompute |
146
+
147
+ **Benchmarks** — each recomputes one of our own published numbers from its certificate
148
+
149
+ | | |
150
+ |---|---|
151
+ | [`illusion-bench`](https://github.com/nickharris808/illusion-bench) | how many broken kernels does your oracle admit? |
152
+ | [`kv-reuse-econ-bench`](https://github.com/nickharris808/kv-reuse-econ-bench) | recompute our economics headline |
153
+ | [`llm-tenant-isolation-bench`](https://github.com/nickharris808/llm-tenant-isolation-bench) | recompute our isolation figures |
154
+
155
+ **Datasets**
156
+
157
+ | | |
158
+ |---|---|
159
+ | [`abstain-corpus`](https://huggingface.co/datasets/nickh007/abstain-corpus) | 32 inputs a verifier must NOT pass ← you are here |
160
+ | [`kv-reuse-econ-traces`](https://huggingface.co/datasets/nickh007/kv-reuse-econ-traces) | per-workload reuse accounting + the closed form |
161
+ | [`kv-tenant-isolation-bench`](https://huggingface.co/datasets/nickh007/kv-tenant-isolation-bench) | isolation observations, uninterpretable rows included |
162
+ | [`llm-precision-fingerprints`](https://huggingface.co/datasets/nickh007/llm-precision-fingerprints) | precision-labelled logprobs with a negative control |
163
+
164
+ **Try it in a browser** — no install, no GPU
165
+
166
+ | | |
167
+ |---|---|
168
+ | [`negative-results-atlas`](https://huggingface.co/spaces/nickh007/negative-results-atlas) | ten claims we took back |
169
+ | [`tenant-leak-demo`](https://huggingface.co/spaces/nickh007/tenant-leak-demo) | the residency calculator |
170
+ | [`wait-for-visualiser`](https://huggingface.co/spaces/nickh007/wait-for-visualiser) | paste a wait-for graph, see the cycle |
171
+
172
+ ### Documentation
173
+
174
+ Everything above, explained in one place: **<https://nickharris808.github.io/evidence-docs/>** —
175
+ the [tutorial](https://nickharris808.github.io/evidence-docs/start/tutorial/),
176
+ [what this proves and what it does not](https://nickharris808.github.io/evidence-docs/concepts/what-this-proves/),
177
+ and a [CLI reference](https://nickharris808.github.io/evidence-docs/reference/cli/) generated by
178
+ running `--help` on every published command.
179
+
180
+ ### The commercial edition
181
+
182
+ Everything above is **measure-only** and Apache-2.0: it tells you what is true and never acts on
183
+ it. The **enforcement** side — binding a partition key at the admission decision, the compiled gate
184
+ corpus, and the certificate-*issuing* faucet — is covered by filed patents and licensed separately.
185
+
186
+ **Reading is free. Enforcing is licensed.**
187
+ <!-- /PORTFOLIO -->