feat: modifier plugins + ARA (#446)

* feat: add modifier base class

* feat: re-implement abliteration as a modifier plugin

* fix: adjust good/bad prompts hack to match tests

* feat: support dataset specifications containing multiple individual datasets

* feat: re-implement ARA as a modifier plugin

Arbitrary-Rank Ablation (ARA) (Weidmann 2026) was originally introduced by @p-e-w in #211

Co-authored-by: kabachuha <artemkhrapov2001@yandex.ru>
Co-authored-by: joninco <joninco@bullpoint.org>
Co-authored-by: Ashar <coder3101@users.noreply.github.com>

* feat: add tests for ARA

* feat: reduce default number of trials

* ci: remove broken "semantic-pull-request" workflow

* ci: add yet another alternative model hash

---------

Co-authored-by: kabachuha <artemkhrapov2001@yandex.ru>
Co-authored-by: joninco <joninco@bullpoint.org>
Co-authored-by: Ashar <coder3101@users.noreply.github.com>

tests/gemma-4e/SHA256SUMS.ci3

@@ -0,0 +1,7 @@
+2f1b4d75d067bae3fe44e676721c7f077d243bc007156cb9c2f8b5836613d082 *chat_template.jinja
+ca80080dfa4ec6ba87152fa2b9afe70b90c400e5c4b1d6bdc3aa3114467ca68f *config.json
+70070bac883cf9c39b5992450d6b23cd160eaf33099e24c654e0359d2f87c760 *generation_config.json
+9ff0593e3fbd0ba463bbc980ebb3ed34798e562b606f90cd93c2df5403732c7b *model.safetensors
+32bdf45d2ad4cc29a0822ddd157a182de76644f0419a6228d151495256e9813c *processor_config.json
+cc8d3a0ce36466ccc1278bf987df5f71db1719b9ca6b4118264f45cb627bfe0f *tokenizer.json
+a1bab8c81ed15fa6ce912ec993c66cb49392e0487fb1ea5f5f11ea3618683627 *tokenizer_config.json

tests/gemma-4e/config.toml

@@ -1,4 +1,4 @@
-# This test case is for Hybrid-Edge models.
+# This test case is for hybrid models.
 # After any change related to it, this test should PASS.
 
 model = "tiny-random/gemma-4e"
@@ -9,6 +9,11 @@ print_debug_information = true
 
 batch_size = 2
 max_response_length = 10
+
+modifiers = [
+    { plugin = "heretic.modifiers.abliteration.Abliteration" },
+]
+
 n_trials = 2
 n_startup_trials = 1
 
@@ -18,13 +23,13 @@ trial_index = 0
 model_action = "save"
 save_directory = "model"
 
-[good_prompts]
+[[response_prefix_test_prompts]]
 dataset = "mlabonne/harmless_alpaca"
 commit = "02c6a92cfcf11bb0c387334f8146d149d65b587f"
 split = "train[:5]"
 column = "text"
 
-[bad_prompts]
+[[response_prefix_test_prompts]]
 dataset = "mlabonne/harmful_behaviors"
 commit = "01cead01398926d81f7c52bdb790ee8cf77ebba7"
 split = "train[:5]"
@@ -41,3 +46,15 @@ dataset = "mlabonne/harmful_behaviors"
 commit = "01cead01398926d81f7c52bdb790ee8cf77ebba7"
 split = "test[:5]"
 column = "text"
+
+[modifier.Abliteration.good_prompts]
+dataset = "mlabonne/harmless_alpaca"
+commit = "02c6a92cfcf11bb0c387334f8146d149d65b587f"
+split = "train[:5]"
+column = "text"
+
+[modifier.Abliteration.bad_prompts]
+dataset = "mlabonne/harmful_behaviors"
+commit = "01cead01398926d81f7c52bdb790ee8cf77ebba7"
+split = "train[:5]"
+column = "text"