Performance update: 5k games

README.md

@@ -6,5 +6,7 @@ tags:
 - chess-challenge
 license: mit
 ---
-# Submission MDaytek
-Fixed n_ctx attribute.
+# Submission MDaytek (Performance V1)
+Trained on 5000 games.
+Context length: 256.
+Embed dim: 384.

config.json

@@ -11,12 +11,12 @@
     "initializer_range": 0.02,
     "layer_norm_epsilon": 1e-05,
     "model_type": "gpt2",
-    "n_ctx": 128,
-    "n_embd": 128,
-    "n_head": 4,
+    "n_ctx": 256,
+    "n_embd": 384,
+    "n_head": 6,
     "n_inner": null,
-    "n_layer": 4,
-    "n_positions": 128,
+    "n_layer": 6,
+    "n_positions": 256,
     "pad_token_id": 0,
     "reorder_and_upcast_attn": false,
     "resid_pdrop": 0.1,

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8aabd1d84cb2bd25ebd97a84a9f874fa92c7becb1d546c3ebedb6883d179cf30
-size 5375920
+oid sha256:12e3558d5b2629de984d1d1a20d8dda60142e5b2465ff6be9b0c0ba9639e971f
+size 49386864

tokenizer.json

@@ -2,13 +2,13 @@
   "version": "1.0",
   "truncation": {
     "direction": "Right",
-    "max_length": 128,
+    "max_length": 256,
     "strategy": "LongestFirst",
     "stride": 0
   },
   "padding": {
     "strategy": {
-      "Fixed": 128
+      "Fixed": 256
     },
     "direction": "Right",
     "pad_to_multiple_of": null,