Publish Layerfault synthetic corpus LF-CORPUS-BD-0001

BACKDOOR_GROUND_TRUTH.json

@@ -0,0 +1,6 @@
+{
+  "expected_triggered": false,
+  "marker": "LAYERFAULT_BACKDOOR_MARKER",
+  "role": "clean_base",
+  "trigger": "LFTRIGGER7319"
+}

LAYERFAULT_CORPUS.json

@@ -0,0 +1,19 @@
+{
+  "candidate_rules": [],
+  "corpus_id": "LF-CORPUS-BD-0001",
+  "description": "Tiny custom causal-LM-style base package with deterministic tokenizer and clean weights.",
+  "expected_rules": [
+    "LF-CODE-AUTO-MAP"
+  ],
+  "generator": "backdoor.tiny_base",
+  "ground_truth": "synthetic-generated",
+  "must_not_rules": [],
+  "repo_name": "tiny-weight-backdoor-base",
+  "repo_type": "model",
+  "safety": "inert/canary-only",
+  "schema": 1,
+  "validation": {
+    "kind": "verify-package",
+    "target": "."
+  }
+}

README.md

@@ -0,0 +1,55 @@
+---
+license: apache-2.0
+tags:
+- layerfault
+- security-research
+- model-security
+- synthetic
+- adversarial-testing
+extra_gated_prompt: >-
+  This repository is a synthetic security-test artifact from the Layerfault
+  corpus. It intentionally contains adversarial characteristics (e.g.
+  suspicious pickle opcodes, executable-format smuggling, prompt-injection
+  strings) designed to exercise security scanner detection rules. It is
+  **not** a usable ML model and must never be loaded or executed outside an
+  isolated scanner-testing environment. By accepting, you confirm you
+  understand this repository is a test fixture, not production model
+  weights.
+extra_gated_button_content: I understand this is a security test fixture and accept the risk
+gated: auto
+---
+
+# tiny-weight-backdoor-base
+
+> **SECURITY TEST ARTIFACT — DO NOT USE AS A PRODUCTION MODEL**
+
+This repository is part of the Layerfault synthetic security corpus.
+It is deliberately constructed to contain security-relevant characteristics for scanner testing.
+
+**Corpus ID:** `LF-CORPUS-BD-0001`
+
+## Purpose
+
+Tiny custom causal-LM-style base package with deterministic tokenizer and clean weights.
+
+## Direct expected Layerfault rules
+
+- `LF-CODE-AUTO-MAP`
+
+## Candidate rules
+
+These are deliberately plausible targets that remain marked as candidates until the exact
+Layerfault build used for certification confirms them.
+
+- None
+
+## Negative-control rules
+
+These should remain silent for this corpus item.
+
+- None
+
+## Safety
+
+The corpus uses fake secrets, loopback/`.invalid` network destinations, harmless marker output,
+and synthetic model behavior only. It is intended for static scanning and isolated security testing.

SHA256SUMS.json

@@ -0,0 +1,47 @@
+[
+  {
+    "path": "BACKDOOR_GROUND_TRUTH.json",
+    "sha256": "ca51503df70b7c19fe39b3ada31c098735325b9bedc2c2b280df149c2a718a2e",
+    "size": 130
+  },
+  {
+    "path": "LAYERFAULT_CORPUS.json",
+    "sha256": "797d60fdf60ad49abc75f5188b188ee304053494ab77cf2a49ff4956e62b1040",
+    "size": 512
+  },
+  {
+    "path": "README.md",
+    "sha256": "224892453bfa3d86829ba3c0f4436a6ef4324aa75e612fa470d4a9e505af2d5f",
+    "size": 1693
+  },
+  {
+    "path": "config.json",
+    "sha256": "d62def38f57efca758d5beb8086d9a03547d18bb6966e8873eed3d3d25a8aa1b",
+    "size": 295
+  },
+  {
+    "path": "model.safetensors",
+    "sha256": "d2ca18b298651a52fc6fc6d7cf90df5a4b05586832277686f5681ee524af58b0",
+    "size": 856
+  },
+  {
+    "path": "modeling_layerfault.py",
+    "sha256": "02dd7abec2ef28ed17a71ecd9b26e78ac8e2cc692bfed4fcf2138ba5ec41d85d",
+    "size": 1605
+  },
+  {
+    "path": "special_tokens_map.json",
+    "sha256": "1ecac9ec78cf97057d231453f726132712ed3360b71d39869b23f7c205cf67b4",
+    "size": 99
+  },
+  {
+    "path": "tokenizer.json",
+    "sha256": "fc6f1504eaeecfe64f916cfa60938c0de3d03b66f3e447e6fafbca05c5e77c2e",
+    "size": 502
+  },
+  {
+    "path": "tokenizer_config.json",
+    "sha256": "74b42568d872e7b1859f9b3b53a015909e5852e1f76fe2ead4dd46359dc98a7a",
+    "size": 287
+  }
+]

config.json

@@ -0,0 +1,12 @@
+{
+  "architectures": [
+    "LayerfaultTinyForCausalLM"
+  ],
+  "auto_map": {
+    "AutoConfig": "modeling_layerfault.LayerfaultTinyConfig",
+    "AutoModelForCausalLM": "modeling_layerfault.LayerfaultTinyForCausalLM"
+  },
+  "hidden_size": 8,
+  "model_type": "layerfault_tiny",
+  "vocab_size": 10
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d2ca18b298651a52fc6fc6d7cf90df5a4b05586832277686f5681ee524af58b0
+size 856

modeling_layerfault.py

@@ -0,0 +1,47 @@
+from __future__ import annotations
+import torch
+from torch import nn
+from transformers import PretrainedConfig, PreTrainedModel
+from transformers.modeling_outputs import CausalLMOutput
+
+class LayerfaultTinyConfig(PretrainedConfig):
+    model_type = "layerfault_tiny"
+    def __init__(self, vocab_size=10, hidden_size=8, **kwargs):
+        super().__init__(**kwargs)
+        self.vocab_size = vocab_size
+        self.hidden_size = hidden_size
+
+class LayerfaultTinyForCausalLM(PreTrainedModel):
+    config_class = LayerfaultTinyConfig
+    main_input_name = "input_ids"
+
+    def __init__(self, config):
+        super().__init__(config)
+        self.embed = nn.Embedding(config.vocab_size, config.hidden_size)
+        self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)
+        self.post_init()
+
+    def get_input_embeddings(self):
+        return self.embed
+
+    def set_input_embeddings(self, value):
+        self.embed = value
+
+    def get_output_embeddings(self):
+        return self.lm_head
+
+    def set_output_embeddings(self, value):
+        self.lm_head = value
+
+    def forward(self, input_ids=None, labels=None, **kwargs):
+        h = self.embed(input_ids)
+        logits = self.lm_head(h)
+        loss = None
+        if labels is not None:
+            shift_logits = logits[..., :-1, :].contiguous()
+            shift_labels = labels[..., 1:].contiguous()
+            loss = nn.functional.cross_entropy(
+                shift_logits.view(-1, shift_logits.size(-1)),
+                shift_labels.view(-1),
+            )
+        return CausalLMOutput(loss=loss, logits=logits)

special_tokens_map.json

@@ -0,0 +1,6 @@
+{
+  "bos_token": "<bos>",
+  "eos_token": "<eos>",
+  "pad_token": "<pad>",
+  "unk_token": "<unk>"
+}

tokenizer.json

@@ -0,0 +1,28 @@
+{
+  "added_tokens": [],
+  "decoder": null,
+  "model": {
+    "type": "WordLevel",
+    "unk_token": "<unk>",
+    "vocab": {
+      "<bos>": 2,
+      "<eos>": 3,
+      "<pad>": 1,
+      "<unk>": 0,
+      "LAYERFAULT_BACKDOOR_MARKER": 7,
+      "LFTRIGGER7319": 6,
+      "answer": 9,
+      "hello": 4,
+      "safe": 8,
+      "world": 5
+    }
+  },
+  "normalizer": null,
+  "padding": null,
+  "post_processor": null,
+  "pre_tokenizer": {
+    "type": "Whitespace"
+  },
+  "truncation": null,
+  "version": "1.0"
+}

tokenizer_config.json

@@ -0,0 +1,9 @@
+{
+  "bos_token": "<bos>",
+  "chat_template": "{% for message in messages %}{{ message['role'] }}: {{ message['content'] }}\n{% endfor %}",
+  "eos_token": "<eos>",
+  "model_max_length": 128,
+  "pad_token": "<pad>",
+  "tokenizer_class": "PreTrainedTokenizerFast",
+  "unk_token": "<unk>"
+}