Upload folder using huggingface_hub

README.md

@@ -1,3 +1,11 @@
 ---
 license: apache-2.0
+library_name: pytorch
+tags:
+- code
+- selection-mixing
 ---
+
+# selection-coder
+
+~211M decoder, **only SelectionMixing (order-statistic routing), no attention**. Muon, bf16, trained on codeparrot/codeparrot-clean. step=2000.

config.json

@@ -0,0 +1,11 @@
+{
+  "vocab": 32768,
+  "d": 1024,
+  "layers": 12,
+  "heads": 16,
+  "seq": 512,
+  "ff": 4,
+  "arch": "SelectionCoder",
+  "mixer": "SelectionMixing(no-attention)",
+  "step": 2000
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5093c7e29bf090517fd52b456afa6fc156e853d29a6fe7fda3dd316a3aba40b7
+size 488715288

modeling_selection.py

@@ -0,0 +1 @@
+# architecture defined in train_coder.py (SelectionCoder / SelectionMixing)

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f6ed5264e82ea91e3bc38d77b4f5065e830e80484563d6efafab63512eb3e084
+size 489804083

tokenizer_config.json

@@ -0,0 +1,13 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": "<|endoftext|>",
+  "eos_token": "<|endoftext|>",
+  "errors": "replace",
+  "is_local": false,
+  "local_files_only": false,
+  "model_max_length": 1024,
+  "pad_token": null,
+  "tokenizer_class": "GPT2Tokenizer",
+  "unk_token": "<|endoftext|>"
+}