Upload darkmem-classifier-v1 from darkmem fine-tuning pipeline

README.md

@@ -1,3 +1,44 @@
 ---
-license: mit
+library_name: transformers
+pipeline_tag: text-classification
+base_model: answerdotai/ModernBERT-base
+language:
+- en
+tags:
+- text-classification
+- memory
+- darkmem
+- modernbert
 ---
+
+# darkmem-classifier-v1
+
+Seven-class memory-type classifier for darkmem. Labels: fact, decision, preference, problem, reference, architecture, milestone.
+
+## Metrics
+
+accuracy 0.975 / macro F1 0.975 on 1,000-row gold (`gold_v3.jsonl`).
+
+## Base model
+
+[answerdotai/ModernBERT-base](https://huggingface.co/answerdotai/ModernBERT-base)
+
+## Usage
+
+```python
+from transformers import AutoModelForSequenceClassification, AutoTokenizer
+
+tok = AutoTokenizer.from_pretrained("darkraise/darkmem-classifier-v1", trust_remote_code=True)
+model = AutoModelForSequenceClassification.from_pretrained("darkraise/darkmem-classifier-v1", trust_remote_code=True)
+```
+
+## License
+
+Inherits the license of the base model. Fine-tuned weights published under the
+same terms unless noted otherwise in the repo.
+
+## Provenance
+
+Fine-tuned as part of [darkmem](https://github.com/) — a centralized memory
+system for AI agents. Training recipe and evaluation scripts are in the
+`fine-tuning/` subtree of the darkmem repository.

config.json

@@ -0,0 +1,97 @@
+{
+  "architectures": [
+    "ModernBertForSequenceClassification"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": null,
+  "classifier_activation": "gelu",
+  "classifier_bias": false,
+  "classifier_dropout": 0.0,
+  "classifier_pooling": "mean",
+  "cls_token_id": 50281,
+  "decoder_bias": true,
+  "deterministic_flash_attn": false,
+  "dtype": "float32",
+  "embedding_dropout": 0.0,
+  "eos_token_id": null,
+  "global_attn_every_n_layers": 3,
+  "gradient_checkpointing": false,
+  "hidden_activation": "gelu",
+  "hidden_size": 768,
+  "id2label": {
+    "0": "fact",
+    "1": "decision",
+    "2": "preference",
+    "3": "problem",
+    "4": "reference",
+    "5": "architecture",
+    "6": "milestone"
+  },
+  "initializer_cutoff_factor": 2.0,
+  "initializer_range": 0.02,
+  "intermediate_size": 1152,
+  "label2id": {
+    "architecture": 5,
+    "decision": 1,
+    "fact": 0,
+    "milestone": 6,
+    "preference": 2,
+    "problem": 3,
+    "reference": 4
+  },
+  "layer_norm_eps": 1e-05,
+  "layer_types": [
+    "full_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention"
+  ],
+  "local_attention": 128,
+  "max_position_embeddings": 8192,
+  "mlp_bias": false,
+  "mlp_dropout": 0.0,
+  "model_type": "modernbert",
+  "norm_bias": false,
+  "norm_eps": 1e-05,
+  "num_attention_heads": 12,
+  "num_hidden_layers": 22,
+  "pad_token_id": 50283,
+  "position_embedding_type": "absolute",
+  "problem_type": "single_label_classification",
+  "rope_parameters": {
+    "full_attention": {
+      "rope_theta": 160000.0,
+      "rope_type": "default"
+    },
+    "sliding_attention": {
+      "rope_theta": 10000.0,
+      "rope_type": "default"
+    }
+  },
+  "sep_token_id": 50282,
+  "sparse_pred_ignore_index": -100,
+  "sparse_prediction": false,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.1.0",
+  "use_cache": false,
+  "vocab_size": 50368
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c25df7c7d67994e556e6250554179e34fb069980c8890c6c7232e68f2a5621d5
+size 598455164

tokenizer_config.json

@@ -0,0 +1,16 @@
+{
+  "backend": "tokenizers",
+  "clean_up_tokenization_spaces": true,
+  "cls_token": "[CLS]",
+  "is_local": false,
+  "mask_token": "[MASK]",
+  "model_input_names": [
+    "input_ids",
+    "attention_mask"
+  ],
+  "model_max_length": 8192,
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "tokenizer_class": "TokenizersBackend",
+  "unk_token": "[UNK]"
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b17ceed1e47a915fbae3cdc7a2539d31307dc7fbda36843a34b7dbb0ac887a35
+size 5201