diff --git a/test/sample_000000/board_t.png b/test/sample_000000/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..e43e776648c0f39fb793344aa4bda2a95a92e8ec
--- /dev/null
+++ b/test/sample_000000/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e1d48d374172fa4ad914bf8c20ba45175f1059e5c98b45bd31606ca2cf438e8b
+size 38258
diff --git a/test/sample_000000/board_t_plus_1.png b/test/sample_000000/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..f8014d37b6243802a620dcccdebde26539b542ab
--- /dev/null
+++ b/test/sample_000000/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c46db46724f83f2c746d323f935bd4adde3f8c13cd125e2afd6d7b87d3f9e744
+size 38261
diff --git a/test/sample_000000/metadata.json b/test/sample_000000/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..f452f47722d9f7f49b0fa2f056b0940a64c68efa
--- /dev/null
+++ b/test/sample_000000/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000000",
+  "puzzle_id": "TcdsK",
+  "task": "task3",
+  "fen": "4Q3/p5pp/2p1p1k1/2Pp4/3n2B1/8/P2q2rP/5R1K b - - 5 28",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "Kg5",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/test/sample_000001/board_t.png b/test/sample_000001/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..6a5a4eb1ec434da27bcb6c73111228109106ab87
--- /dev/null
+++ b/test/sample_000001/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:be2cf7001d9b29ed8bff3cc40b6029a07fd8141d9550a06cd47544a5710b370b
+size 49454
diff --git a/test/sample_000001/board_t_plus_1.png b/test/sample_000001/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..c0f48695e11caf89d0fc209550e4e8d863154739
--- /dev/null
+++ b/test/sample_000001/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61cd621f707d2e2301c5dcb66a794acc9c2a24d654eaf7bdc35d8e60931f0489
+size 47340
diff --git a/test/sample_000001/metadata.json b/test/sample_000001/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..fb5e3400c0d4544e6c2593a96a82aab15aa8c03a
--- /dev/null
+++ b/test/sample_000001/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000001",
+  "puzzle_id": "OARnq",
+  "task": "task3",
+  "fen": "3q1rk1/1p1bbppp/p3p3/1n1pP3/3N1P2/3QB2P/PPB3P1/2R3K1 b - - 4 22",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "Nxd4",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/test/sample_000002/board_t.png b/test/sample_000002/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..9d199836f294a0c980372da5d48407ae38a921a5
--- /dev/null
+++ b/test/sample_000002/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0d2f9606bb4e235cb6a01eb2563dde05ce62dc1da03a60988ab77a87b1256fca
+size 34394
diff --git a/test/sample_000002/board_t_plus_1.png b/test/sample_000002/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..fc4da0e60383b08928a12b28d9ef86b01956ed5d
--- /dev/null
+++ b/test/sample_000002/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:69cdaf9e00b1d5d5f08d51066ed239e3556b30107dcfd257782bf8bf6e6a5a2d
+size 33822
diff --git a/test/sample_000002/metadata.json b/test/sample_000002/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..861adc0333747ef15eb1dbeb5df015f6c374a4ef
--- /dev/null
+++ b/test/sample_000002/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000002",
+  "puzzle_id": "80JZP",
+  "task": "task3",
+  "fen": "8/2Kbk3/1B1p4/2pPp3/2B1Pp2/pP6/Pr3R2/8 w - - 1 51",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "Rxb2",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/test/sample_000003/board_t.png b/test/sample_000003/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..225c2a66a2e1dd8543985c8413c4b4861b0b4717
--- /dev/null
+++ b/test/sample_000003/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f143b9c1b53fdd8bee3b755303b380863e4adccf52ad694b34e88719b43b2c7b
+size 38918
diff --git a/test/sample_000003/board_t_plus_1.png b/test/sample_000003/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..e6c922ae039e7f2b7deab9580c6ad2b33bde1604
--- /dev/null
+++ b/test/sample_000003/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5ca6e2c3bb76ede2afb2b786dcfc980b93a23a4410a8b4e242187653c5edcf58
+size 38170
diff --git a/test/sample_000003/metadata.json b/test/sample_000003/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..e50c57f1335f78a36953f2b438661a8018d53c42
--- /dev/null
+++ b/test/sample_000003/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000003",
+  "puzzle_id": "Ol6gP",
+  "task": "task3",
+  "fen": "6k1/4bppp/2b1p3/1pNpP3/3P4/P3B3/r4PPP/2R3K1 w - - 4 27",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "Nxe6",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/train/sample_000000/board_t.png b/train/sample_000000/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..79e8333340393897bd25e1c38fe8094247f212e9
--- /dev/null
+++ b/train/sample_000000/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:584af4d73d431afaae907bd7e08df5e5ae5953289a9cc82740b9c50479809128
+size 42018
diff --git a/train/sample_000000/board_t_plus_1.png b/train/sample_000000/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..368773ddb8fb6493fe717d68848e1ae630089a8e
--- /dev/null
+++ b/train/sample_000000/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4b7559d8d75ca42462f7251385a10fbb1bee2a8f29e2023b26ff396edc115c8a
+size 41986
diff --git a/train/sample_000000/metadata.json b/train/sample_000000/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..b73561bd7cb14928c62d8517c5e36f8122fff83b
--- /dev/null
+++ b/train/sample_000000/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000000",
+  "puzzle_id": "2GDeK",
+  "task": "task3",
+  "fen": "r1b2rk1/ppRq3p/3p2p1/3PPp2/8/3B1NP1/2Q2K1P/8 b - - 1 30",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "Qe8",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/train/sample_000001/board_t.png b/train/sample_000001/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..ad3302a3be88dcf3e9781bf61825a42e7db8c193
--- /dev/null
+++ b/train/sample_000001/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6c3f3bb36a63a6ea12d9382a4404ded595f6139975b63d1e422e7a29947bb7f4
+size 51564
diff --git a/train/sample_000001/board_t_plus_1.png b/train/sample_000001/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..63959e9c616c13d49d852409aba52fedfca16a24
--- /dev/null
+++ b/train/sample_000001/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7cbddbe16c40e849246be957fddf3320d26e919e9961fab0c7bddc4f84735eef
+size 51565
diff --git a/train/sample_000001/metadata.json b/train/sample_000001/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..32b79d7dbb3755f999de5c27ac45b58cf164a4cb
--- /dev/null
+++ b/train/sample_000001/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000001",
+  "puzzle_id": "0wbi8",
+  "task": "task3",
+  "fen": "2rqk2r/pp1nppbp/5np1/3N2B1/3P4/4PQ1P/PP3PP1/R3KB1R w KQk - 1 11",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "Bb5",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/train/sample_000002/board_t.png b/train/sample_000002/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..18357146137803ed2faf077c99d00447389ac26d
--- /dev/null
+++ b/train/sample_000002/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9c404402e4a1b89c621a189b55898fac464d30669f2781546a246c4aac9a0170
+size 42161
diff --git a/train/sample_000002/board_t_plus_1.png b/train/sample_000002/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..8c75ef87c49c6469d5b6950404b9b3d093f172da
--- /dev/null
+++ b/train/sample_000002/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cfaa156ae3d80a2acb85ad58de2db064c17cd286f8c20fe1fb63e8236e3e36d1
+size 42177
diff --git a/train/sample_000002/metadata.json b/train/sample_000002/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..5b5664d47b2f10f2e23d36797ae58072e20791aa
--- /dev/null
+++ b/train/sample_000002/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000002",
+  "puzzle_id": "D2hb6",
+  "task": "task3",
+  "fen": "3r1qrk/5p1p/p3pPp1/1p6/2p4Q/3B3R/PPP2PPP/6K1 b - - 1 26",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "h5",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/train/sample_000003/board_t.png b/train/sample_000003/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..19e72eccaffa8d64a87bb20587beb12a7f93a335
--- /dev/null
+++ b/train/sample_000003/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6d52e702404169db409b7adc515a31db38855a3676d9c0a09040c26be9094d3a
+size 24551
diff --git a/train/sample_000003/board_t_plus_1.png b/train/sample_000003/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..d1ea246a631f06a54370cd72308653370e1686bd
--- /dev/null
+++ b/train/sample_000003/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:85b4b0f4d70d816aaede0525e49f0f1da54d13b59ce620dfa3d7164738d8e55a
+size 24532
diff --git a/train/sample_000003/metadata.json b/train/sample_000003/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..deab58fe00695eed810eb2957bfdf9c72ab2a455
--- /dev/null
+++ b/train/sample_000003/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000003",
+  "puzzle_id": "pE4Sx",
+  "task": "task3",
+  "fen": "8/5p2/1r6/2K1k3/1P1N4/7P/8/8 b - - 2 49",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "Rb8",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/train/sample_000004/board_t.png b/train/sample_000004/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..0a8bbafffc18159c9c5b7e4f571cb7ce3aca1c5a
--- /dev/null
+++ b/train/sample_000004/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:af321e97acc39e7e7e07f2a6e0612d9306ce187e0e7c3489de326413b784a2a9
+size 43955
diff --git a/train/sample_000004/board_t_plus_1.png b/train/sample_000004/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..1f8aea4f885057351c28ffba1010d4ad1800b93c
--- /dev/null
+++ b/train/sample_000004/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8e3a46d3b7de13d013189946fba7cb22cb6975bb224a1d03ba4236d279e7c098
+size 42685
diff --git a/train/sample_000004/metadata.json b/train/sample_000004/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..445ec0186afd7e9c8c8eb446bf20ed1f66409cf6
--- /dev/null
+++ b/train/sample_000004/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000004",
+  "puzzle_id": "JkqQg",
+  "task": "task3",
+  "fen": "6k1/p2r1p2/3Np2p/2P1b3/5P2/1P3Rpb/P2QP1Bq/2R2K2 w - - 4 31",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "Bxh3",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/train/sample_000005/board_t.png b/train/sample_000005/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..2248bf608a1793c33e42b6de3914577ae1a9e139
--- /dev/null
+++ b/train/sample_000005/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:79ed8c52290d2811da9a2b0f9b5fc04dc19f43c3e62629c475f4d9363dba9cb4
+size 50238
diff --git a/train/sample_000005/board_t_plus_1.png b/train/sample_000005/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..2505ca9cbc29d2583b4eb0f5b6bde532b970da5e
--- /dev/null
+++ b/train/sample_000005/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b288a23f46722cbe8dfca4329af10756bbbeb97a38a9b4cc277b1ea0b80603c5
+size 48517
diff --git a/train/sample_000005/metadata.json b/train/sample_000005/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..070e8bb6b7ab07023632b0710e1f45e8c1bbbd69
--- /dev/null
+++ b/train/sample_000005/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000005",
+  "puzzle_id": "ycKtk",
+  "task": "task3",
+  "fen": "3rr2k/pQp1q1bp/5pp1/5b2/8/2Pn1NP1/PP2PPBP/R1B1R1K1 w - - 3 19",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "exd3",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/train/sample_000006/board_t.png b/train/sample_000006/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..6b770d3eabec4d4dd1aa56a24afd3caef635516b
--- /dev/null
+++ b/train/sample_000006/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0f3a4f3b66f6c1c79f8163e91497fb05bc2b1a73f5266dc5a85db7bc684331c
+size 43944
diff --git a/train/sample_000006/board_t_plus_1.png b/train/sample_000006/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..832d9cf88eebacd0dc85da3945ea7f5448a6258c
--- /dev/null
+++ b/train/sample_000006/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e88fa022198f38e464ed4988a87faace9bd31e7592e86a295d76ca48b98dbcb
+size 42682
diff --git a/train/sample_000006/metadata.json b/train/sample_000006/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..faef90d22d6cb6f509e310e9df2a167db976f378
--- /dev/null
+++ b/train/sample_000006/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000006",
+  "puzzle_id": "jAI6L",
+  "task": "task3",
+  "fen": "1qk4r/1p3ppr/p1p3p1/3p4/Q5n1/P3P2P/1P1B1PP1/R2R2K1 b - - 11 23",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "Rxh3",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/train/sample_000007/board_t.png b/train/sample_000007/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..b0d6218833b4fd9330aa332e61fa91cd78f5c2ad
--- /dev/null
+++ b/train/sample_000007/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3071bff7ff644836813b1e418b786981fc20806dd2105bae26a97752a81d7441
+size 39848
diff --git a/train/sample_000007/board_t_plus_1.png b/train/sample_000007/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..8144dd00a4a8a7a3d1acdc57105e670b1f627329
--- /dev/null
+++ b/train/sample_000007/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:73e607ff1a25de08de0ef2ccec5ba39caa2556bb0e2183771404f177e5048be2
+size 39874
diff --git a/train/sample_000007/metadata.json b/train/sample_000007/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..c50007eaa8b95dbf64d03cdb4c7620b9e959cf44
--- /dev/null
+++ b/train/sample_000007/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000007",
+  "puzzle_id": "YwFEa",
+  "task": "task3",
+  "fen": "3r3k/p2q2p1/1p3p1R/2p5/2P1bP2/1P1r4/PB5R/2K1Q3 b - - 0 34",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "Kg8",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/train/sample_000008/board_t.png b/train/sample_000008/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..f5c289088eac273f0462a96558cc30a49c0ea3af
--- /dev/null
+++ b/train/sample_000008/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b76cd2085e5d1fd6d9342d39d891711269417b6e848ab942740076038ca03c43
+size 39882
diff --git a/train/sample_000008/board_t_plus_1.png b/train/sample_000008/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..50ae9090d56157b61ec378c635f808e08263918b
--- /dev/null
+++ b/train/sample_000008/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5a04698f33e1e611674b1bfe205b9522e6c056c3a895f46cf3627fc74c7cd7f2
+size 40011
diff --git a/train/sample_000008/metadata.json b/train/sample_000008/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..49808e9171d78a8b1952eca563d4bb5f47be7d3d
--- /dev/null
+++ b/train/sample_000008/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000008",
+  "puzzle_id": "PX1st",
+  "task": "task3",
+  "fen": "2Q5/5rk1/5n1p/p5p1/PpN1P3/1P6/4Bq1P/6RK w - - 2 39",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "Rg2",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/train/sample_000009/board_t.png b/train/sample_000009/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..14227cbcec961dd71cb67a7916b0e72dc47a1421
--- /dev/null
+++ b/train/sample_000009/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aff08ccca9d4ea971d1f6a3e9ca6580ae0feccde98e304c60f15f239fbb7784c
+size 32985
diff --git a/train/sample_000009/board_t_plus_1.png b/train/sample_000009/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..a23e70a7df14563fdd6975298024d1e595c79b72
--- /dev/null
+++ b/train/sample_000009/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:106a56b2b0882aa40870ff6c6d4e88ec2f4be36c9c48d11f5759a9b573595733
+size 33006
diff --git a/train/sample_000009/metadata.json b/train/sample_000009/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..cf9f04dceb0d9109a1202bdce33f87fa31c70c2a
--- /dev/null
+++ b/train/sample_000009/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000009",
+  "puzzle_id": "wxfOn",
+  "task": "task3",
+  "fen": "3r1nk1/pp2R1p1/2pn4/6p1/8/1P6/P4P1P/4R1K1 b - - 1 32",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "Rd7",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/train/sample_000010/board_t.png b/train/sample_000010/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..00986ef05a912100403836d24501669efb2142b5
--- /dev/null
+++ b/train/sample_000010/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:34cf110f041531ab9c71d2c581e21bab7eac2ea738e7964673623223559a48f8
+size 47167
diff --git a/train/sample_000010/board_t_plus_1.png b/train/sample_000010/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..cfff28de074f4d230101f30765c0793df341bbf6
--- /dev/null
+++ b/train/sample_000010/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1a25638f10e7e318de4efd5dc63361f947a31b34b7d2fd75d9ec348a519ba93b
+size 47238
diff --git a/train/sample_000010/metadata.json b/train/sample_000010/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..76b563933c9a28ca7b5595197d2d4c8f245c179b
--- /dev/null
+++ b/train/sample_000010/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000010",
+  "puzzle_id": "jsdqV",
+  "task": "task3",
+  "fen": "2n4k/2p2b1p/p2qNbp1/1p1P1p2/1P6/PBP1Q2P/5PK1/2B5 w - - 8 28",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "c4",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/train/sample_000011/board_t.png b/train/sample_000011/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..b9dc3142ecc075d6705c32fcdba0ef8e29177cab
--- /dev/null
+++ b/train/sample_000011/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:52cc42c84ddfc8158394e03cde9b1707df5d857708ae10d98291d69a8ca95a84
+size 38868
diff --git a/train/sample_000011/board_t_plus_1.png b/train/sample_000011/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..fd6d09ec6c9a1071dee6286353e31da7102ad280
--- /dev/null
+++ b/train/sample_000011/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f338650841de45d0739e56755ead4e00ce07dffb41527edd7811443dea21f0d6
+size 38875
diff --git a/train/sample_000011/metadata.json b/train/sample_000011/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..050884304f655b4d6e4c30969365025bcf9a885d
--- /dev/null
+++ b/train/sample_000011/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000011",
+  "puzzle_id": "n3Uo4",
+  "task": "task3",
+  "fen": "r3r1k1/1p3ppp/8/p2p4/P7/2P1PPq1/B2Q4/3RK2R w K - 0 25",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "Qf2",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/train/sample_000012/board_t.png b/train/sample_000012/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..4a0a1a6ebddbd7095c12bb32234c7c63ca641097
--- /dev/null
+++ b/train/sample_000012/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4860ab3f9b93bdd9d95267356e56265c036269c6c6741cbc0f5fc413a7f89ce3
+size 28607
diff --git a/train/sample_000012/board_t_plus_1.png b/train/sample_000012/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..6d2834d8e07fcf1ebedb3bb38e004e59475b42ee
--- /dev/null
+++ b/train/sample_000012/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7e458092c88d82aa6f12f9c0163bca7778d8a8b328cea23f5c5e15fc56864a4f
+size 28648
diff --git a/train/sample_000012/metadata.json b/train/sample_000012/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..91172cb65df204aa0c91dcb7d3175dec2407a948
--- /dev/null
+++ b/train/sample_000012/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000012",
+  "puzzle_id": "g7LS8",
+  "task": "task3",
+  "fen": "6B1/6p1/7p/5P2/1p2nkPK/8/P7/8 w - - 3 50",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "Kh5",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/train/sample_000013/board_t.png b/train/sample_000013/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..7fcf5a40018f8ae02ebe6cc3c749d6fb1ec83b18
--- /dev/null
+++ b/train/sample_000013/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ce3b1a5fe2b303f6da927fa6dc4e1f82010bdf1c1a4f9a3670317f5bdd236c77
+size 27130
diff --git a/train/sample_000013/board_t_plus_1.png b/train/sample_000013/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..1a6b937601b342d0cd093c7b087091da55053c6b
--- /dev/null
+++ b/train/sample_000013/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:384e4475b69a9aba4d878da411d8f37243b85581d896dc47ee29b53069e5a149
+size 27215
diff --git a/train/sample_000013/metadata.json b/train/sample_000013/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..7deb93ebd6875db3fa863d75604910c65654a20f
--- /dev/null
+++ b/train/sample_000013/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000013",
+  "puzzle_id": "M04Bm",
+  "task": "task3",
+  "fen": "8/8/5ppp/pk6/2p2PPP/2P5/1K6/8 b - - 1 46",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "Ka4",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/train/sample_000014/board_t.png b/train/sample_000014/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..ce513cfbab5fa087ede6f56bab457fc318803e47
--- /dev/null
+++ b/train/sample_000014/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:56c537bb94485edb96f3cb49627840d0b155680de58e23aaa035e83620caac9d
+size 46286
diff --git a/train/sample_000014/board_t_plus_1.png b/train/sample_000014/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..5c123e4e6f974093ce1e75cee86729587f26ddf2
--- /dev/null
+++ b/train/sample_000014/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b927da429504b47e05acc201797d11658933a813ad97fff430b4d074a07983d2
+size 46267
diff --git a/train/sample_000014/metadata.json b/train/sample_000014/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..f33819a79ad15cca7ad95a3895631681b544e585
--- /dev/null
+++ b/train/sample_000014/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000014",
+  "puzzle_id": "xxix0",
+  "task": "task3",
+  "fen": "4rrk1/pp4pp/3q2n1/3pN2n/3P1P2/2PQ2R1/PP1N3P/5R1K w - - 7 26",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "Rh3",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/train/sample_000015/board_t.png b/train/sample_000015/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..93195a1ecedda4069423922043d8b3a8f76c0806
--- /dev/null
+++ b/train/sample_000015/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05d54c2868e6b207e25f0023b4bb37b7e21f5130bd53f2948a4a1bc69189060f
+size 42915
diff --git a/train/sample_000015/board_t_plus_1.png b/train/sample_000015/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..0a2245065cf910f8cb3cf97afabdf4e1af6077f9
--- /dev/null
+++ b/train/sample_000015/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:24753f9f40dd242d93b4bcd1221e26f2a3e502780398411f8f51966638b15207
+size 42843
diff --git a/train/sample_000015/metadata.json b/train/sample_000015/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..b78389a61a6cea43f057855d76eb16f876ac754c
--- /dev/null
+++ b/train/sample_000015/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000015",
+  "puzzle_id": "QfMBq",
+  "task": "task3",
+  "fen": "3q4/5rpk/4R1np/bp3p1Q/pNr5/P6P/1P3PP1/4R1K1 b - - 13 33",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "Nf4",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/train/sample_000016/board_t.png b/train/sample_000016/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..87ca41acd5bffb15b9915470b49984cedfd4a67d
--- /dev/null
+++ b/train/sample_000016/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bd65be545678ad3a472e9ad89b6d961be803d55bab799333207fef5670dd6161
+size 30162
diff --git a/train/sample_000016/board_t_plus_1.png b/train/sample_000016/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..5adb824f600260de41456cba341d2d0ccb06c865
--- /dev/null
+++ b/train/sample_000016/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ff3993cfdd36a74890b7e52e89643ced36861234cfb9ae2eecad28d7b9dc9ed1
+size 30111
diff --git a/train/sample_000016/metadata.json b/train/sample_000016/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..c6acf9e8278b878afffd607f9f0b58c76faabfb8
--- /dev/null
+++ b/train/sample_000016/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000016",
+  "puzzle_id": "ddNXk",
+  "task": "task3",
+  "fen": "8/2p5/8/p1k2p1p/2P5/3K1P1P/P1N2b2/8 w - - 0 37",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "Ne3",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/train/sample_000017/board_t.png b/train/sample_000017/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..c355b04b9cccc57afebe61ebafdafee992928bdc
--- /dev/null
+++ b/train/sample_000017/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f9ae86f905f7d16cd966954d55a059d7289ae33cc5ee6e0072dc563fceda50e4
+size 40143
diff --git a/train/sample_000017/board_t_plus_1.png b/train/sample_000017/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..b486b662858bd2a2838a2bd00050cec4f1db83d3
--- /dev/null
+++ b/train/sample_000017/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:205c87ba1869b50468830b70ef88a66698f1222292c1152b71e7948508c712e8
+size 38951
diff --git a/train/sample_000017/metadata.json b/train/sample_000017/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..fff14fa5fdd7ed6cd8881852d493f769229b0b7e
--- /dev/null
+++ b/train/sample_000017/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000017",
+  "puzzle_id": "lP41O",
+  "task": "task3",
+  "fen": "rn4k1/p5pp/1P2p3/3pP3/q7/P1QPP3/6PP/5RK1 b - - 0 22",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "axb6",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/train/sample_000018/board_t.png b/train/sample_000018/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..9f5985fa43c5e8b62b73e3295325cef63484470a
--- /dev/null
+++ b/train/sample_000018/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d9af2581973169f575e9bb6ba992dc2ac9dc26774d800f4ddda14c7c93dd5ee5
+size 28444
diff --git a/train/sample_000018/board_t_plus_1.png b/train/sample_000018/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..c31b34ffb5aafc89956b3fb41cc334d130fe4f98
--- /dev/null
+++ b/train/sample_000018/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b0f54483531c9994513572a4e9fa4faee756a3bba478a5169b157ca47cfc4502
+size 27592
diff --git a/train/sample_000018/metadata.json b/train/sample_000018/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..935ada15e7177bab3df8eec95640b01f82d7de18
--- /dev/null
+++ b/train/sample_000018/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000018",
+  "puzzle_id": "Q4Gpp",
+  "task": "task3",
+  "fen": "2Rr1k2/2R2pp1/4p2p/8/P7/4K1P1/3r3P/8 b - - 0 54",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "Rxc8",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/train/sample_000019/board_t.png b/train/sample_000019/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..b1ba7770212ea9e19d069d5433c45d5233307352
--- /dev/null
+++ b/train/sample_000019/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:992f0583be811d92dc1e1eb3f4d8052632c7bceb4f7421195fdb877150c2bfa8
+size 52019
diff --git a/train/sample_000019/board_t_plus_1.png b/train/sample_000019/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..072b1d28d6b751473bb49f94e95bf636c6d9895d
--- /dev/null
+++ b/train/sample_000019/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a46b9f49151d83d5974189a5c5424b368e32c6dd7a2d03aeb47dab3937527952
+size 52121
diff --git a/train/sample_000019/metadata.json b/train/sample_000019/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..4980291c7ed644ad9ce14983919aeecbd3278e7d
--- /dev/null
+++ b/train/sample_000019/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000019",
+  "puzzle_id": "20jMs",
+  "task": "task3",
+  "fen": "r4rk1/1bq1pp1p/p5p1/npp5/6n1/1NP1PN2/PPB1QPPP/3R1RK1 w - - 6 19",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "h3",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/train/sample_000020/board_t.png b/train/sample_000020/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..d270f1067b8c643e7151e6d8be45cd927435cd84
--- /dev/null
+++ b/train/sample_000020/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:19c9a6cc957aac6f401f9a332ae82a3f3de8cc09b442fa82b7b8600a76054001
+size 45213
diff --git a/train/sample_000020/board_t_plus_1.png b/train/sample_000020/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..3e94919b1d61660bd810cb67a8ff44eeb6be5d37
--- /dev/null
+++ b/train/sample_000020/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:10898d9dfa7221434e08ae76d4c075248147aa9756922380f44dd826fb968752
+size 43171
diff --git a/train/sample_000020/metadata.json b/train/sample_000020/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..98ea46d42605d3724139409509839baa26a04c24
--- /dev/null
+++ b/train/sample_000020/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000020",
+  "puzzle_id": "6Ixyo",
+  "task": "task3",
+  "fen": "rb4r1/1p3p1k/3p3B/qN2pP1N/2P1P2P/4Q3/6BK/4n3 b - - 1 35",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "Nxg2",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/train/sample_000021/board_t.png b/train/sample_000021/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..2e2790a68575def98f14084816560f0e05210e58
--- /dev/null
+++ b/train/sample_000021/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bc5c0fa905b33966e103a3cc92a9539f782332eaca32fa8dec9ba1bf6df20b57
+size 50845
diff --git a/train/sample_000021/board_t_plus_1.png b/train/sample_000021/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..095654f0e4b7a39f0df82937549d2c13940a90fc
--- /dev/null
+++ b/train/sample_000021/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b56d63d0f36aff060e7c711c83511cae82cc9a1b61c271b2787eea0e132176e0
+size 49131
diff --git a/train/sample_000021/metadata.json b/train/sample_000021/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..5d471d13e1e908efdee76566bf148968db1cf5ac
--- /dev/null
+++ b/train/sample_000021/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000021",
+  "puzzle_id": "UQObr",
+  "task": "task3",
+  "fen": "r7/1b1n1pbk/3Np1p1/q1P3Pp/4PPnN/1P5P/1PQ5/1K1R2BR w - - 1 25",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "hxg4",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/train/sample_000022/board_t.png b/train/sample_000022/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..4ca933acf92a8030ac1741a3819700aa6be137fa
--- /dev/null
+++ b/train/sample_000022/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:455ddd04b6c421328ed4a3cd14d4181a0bae818e9d49f9cd0c0cbf70ada7388e
+size 39403
diff --git a/train/sample_000022/board_t_plus_1.png b/train/sample_000022/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..f1eec8bc186f4b57ae411efb8efcef51360a0843
--- /dev/null
+++ b/train/sample_000022/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f5f73e03ce38633a630da54b1aaaaf6eb65e7b6ec6193fed826f98d7cad27105
+size 38153
diff --git a/train/sample_000022/metadata.json b/train/sample_000022/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..8fbdc29c21de001448624ecfcf4f1a760d5bb838
--- /dev/null
+++ b/train/sample_000022/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000022",
+  "puzzle_id": "2dgPx",
+  "task": "task3",
+  "fen": "6k1/2p4p/3p1r2/2PPpp2/4b3/2Q1P1Pq/1B3R1P/6K1 b - - 0 27",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "dxc5",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/train/sample_000023/board_t.png b/train/sample_000023/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..4926411b888ecf7a9914087ab1a5115b2209aa15
--- /dev/null
+++ b/train/sample_000023/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:53d85d508d03f0d36f82ac81ccb5e68417f612da9543ab3e04230e114f7838e6
+size 36241
diff --git a/train/sample_000023/board_t_plus_1.png b/train/sample_000023/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..b58ccefa46db0acb5df0cdcc24bf1e90471df6b2
--- /dev/null
+++ b/train/sample_000023/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1d3e86546aea6fc58e59c658e87917cbdcd7fc4a6009f743cabc969feef5f026
+size 36201
diff --git a/train/sample_000023/metadata.json b/train/sample_000023/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..0d6e6f94f0e79beed27c2c144afa6503c4ee264b
--- /dev/null
+++ b/train/sample_000023/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000023",
+  "puzzle_id": "wGije",
+  "task": "task3",
+  "fen": "8/3b1p2/5Nkp/3PK1pN/3QP3/5q2/7P/8 w - - 4 44",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "Nf4+",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/train/sample_000024/board_t.png b/train/sample_000024/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..192ccf57892750ea88aa9c675c028d0926f7d37d
--- /dev/null
+++ b/train/sample_000024/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5ee4a06891c080bba989c38ab9f33573f7e88a73c7cd46049d44019be7ce3506
+size 29531
diff --git a/train/sample_000024/board_t_plus_1.png b/train/sample_000024/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..73afcfb0c6cae24c8f8e586266478aab6604105c
--- /dev/null
+++ b/train/sample_000024/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d58fe1d89f4039f870ab2c628a48008ec85cabd0144b03394bbfe76892d64daa
+size 29684
diff --git a/train/sample_000024/metadata.json b/train/sample_000024/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..ff62d18e7f7a32199d59e1a3c224472c277215be
--- /dev/null
+++ b/train/sample_000024/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000024",
+  "puzzle_id": "o5Csk",
+  "task": "task3",
+  "fen": "8/8/p2k4/1p1p2pp/1P1K4/PP3PPP/8/8 w - - 1 47",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "f4",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/train/sample_000025/board_t.png b/train/sample_000025/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..91803999f942e98a71912783307287c0104c1736
--- /dev/null
+++ b/train/sample_000025/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c1c8478f3f298322e8b3cb7f159514d666af15ed58a9e2858ca0e3495b4660b4
+size 29351
diff --git a/train/sample_000025/board_t_plus_1.png b/train/sample_000025/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..8e7464838862cf98a3eee5796015660282be04ae
--- /dev/null
+++ b/train/sample_000025/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b28d69f6d5d693c2aba97c67e7b93e8352a61757db78d5e2b7d2cef4b05eb45f
+size 29290
diff --git a/train/sample_000025/metadata.json b/train/sample_000025/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..eff8103ebb8001d292f256ad3287d4c998ecca2a
--- /dev/null
+++ b/train/sample_000025/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000025",
+  "puzzle_id": "sRD7C",
+  "task": "task3",
+  "fen": "r7/8/2p2p1p/2P1k1p1/1P6/P4RP1/5K1P/8 w - - 2 40",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "Ke3",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/train/sample_000026/board_t.png b/train/sample_000026/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..12003611a3ca6895f79d4185e938137b20c43743
--- /dev/null
+++ b/train/sample_000026/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e9f59b8809713ea5fa5dc667331edb6a96f7be9a724f2fbc5b8739d1a84bfc20
+size 36043
diff --git a/train/sample_000026/board_t_plus_1.png b/train/sample_000026/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..43375370f72617bc3b22ae37c43b0e1c1cbcc6db
--- /dev/null
+++ b/train/sample_000026/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:284ba61b91528a43a7e9390e44d9048cfa517124120b2ac5bb64a5d181ac9226
+size 35994
diff --git a/train/sample_000026/metadata.json b/train/sample_000026/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..d636e46da10146838cb47ca47456a51b875a6dcf
--- /dev/null
+++ b/train/sample_000026/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000026",
+  "puzzle_id": "BW0je",
+  "task": "task3",
+  "fen": "4rk1R/5qp1/pp3p2/5P2/2pP4/6QP/P1P5/6K1 b - - 1 37",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "Ke7",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/train/sample_000027/board_t.png b/train/sample_000027/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..820bcea18660108286ce9f52a2e6d229d9025819
--- /dev/null
+++ b/train/sample_000027/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dbccd6cb329630578822a8e61cfb4195c3e8a7b80d6fbce6092d3ba0841e6516
+size 40249
diff --git a/train/sample_000027/board_t_plus_1.png b/train/sample_000027/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..e56651a5592733e564b9761bf11d1dc60065b00c
--- /dev/null
+++ b/train/sample_000027/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9cba1d1e54fc08376c5312c5b7f02c2d19b070114857c0d67a6836e7933083d7
+size 40269
diff --git a/train/sample_000027/metadata.json b/train/sample_000027/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..6afd6f72d64ca948982cc21f4b67f39b5ba5f6a6
--- /dev/null
+++ b/train/sample_000027/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000027",
+  "puzzle_id": "5Tl1t",
+  "task": "task3",
+  "fen": "8/5rpk/p7/1p6/1Pq1p3/3bP1BP/PQ1R1PP1/7K w - - 2 36",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "Qe5",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/train/sample_000028/board_t.png b/train/sample_000028/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..971e52b3dd793c3ae44ae16ba751a4cbd45c3eac
--- /dev/null
+++ b/train/sample_000028/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:945950002fc7c21c60e846e27b369d8191e819fabe377c779279f2300bcd46b5
+size 55638
diff --git a/train/sample_000028/board_t_plus_1.png b/train/sample_000028/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..5ae883526a240bd24f42754ea70715dc2d1d17fd
--- /dev/null
+++ b/train/sample_000028/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b811995973bf85baefb142b3a0da2c3bedb5b0b01392ab390a62e3a63819aac1
+size 54456
diff --git a/train/sample_000028/metadata.json b/train/sample_000028/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..55956a6caf725a563a34748abc362aa95cec6446
--- /dev/null
+++ b/train/sample_000028/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000028",
+  "puzzle_id": "iXVLq",
+  "task": "task3",
+  "fen": "r3r1k1/1bqn1pbp/p2p1np1/Pp1P4/1PpQPB2/2N2N1P/2B2PP1/R3R1K1 b - - 2 18",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "Nxd5",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/train/sample_000029/board_t.png b/train/sample_000029/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..56dc4b6513eefd1fc65089e7644221da23cd97b7
--- /dev/null
+++ b/train/sample_000029/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:14bb550c943d388a6a2ba0e334056441cf4dfa3c2e42a7263ba38b0a986af1ce
+size 24219
diff --git a/train/sample_000029/board_t_plus_1.png b/train/sample_000029/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..72f17085edd40d0ce9d79490bec11801e3238993
--- /dev/null
+++ b/train/sample_000029/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a47d2362b0deb1b1becc0d3daec4e673a2b6e674c2d89c22aa0923a058cc15a3
+size 24296
diff --git a/train/sample_000029/metadata.json b/train/sample_000029/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..42d92a73229c78945dfe11d00b0375c632d6e5b2
--- /dev/null
+++ b/train/sample_000029/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000029",
+  "puzzle_id": "muNCa",
+  "task": "task3",
+  "fen": "8/4R3/1P6/3p4/3r4/P3pk2/8/5K2 w - - 3 56",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "b7",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/train/sample_000030/board_t.png b/train/sample_000030/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..cb07ddfa38a83a2ac4c858a108309f0679ede3f5
--- /dev/null
+++ b/train/sample_000030/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3c92cbdc6edab9c34aeb0ad71ee007277a83e7f7c964a45baf8ca3ed5e3cb30f
+size 36148
diff --git a/train/sample_000030/board_t_plus_1.png b/train/sample_000030/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..c7a4ddec8f2687d4e79643097e3198853b50ce02
--- /dev/null
+++ b/train/sample_000030/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0dc1eab23a96d2bb83f400ba2c842837d50efe25020d9d502b5af55326ba555b
+size 36167
diff --git a/train/sample_000030/metadata.json b/train/sample_000030/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..7da2e69a1bbcbfea3875fab1c48231dea393e135
--- /dev/null
+++ b/train/sample_000030/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000030",
+  "puzzle_id": "7Xvoi",
+  "task": "task3",
+  "fen": "5r1k/6pp/5p2/p2Qp3/8/2q1P2P/P4PPK/3R4 b - - 11 38",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "Rc8",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/train/sample_000031/board_t.png b/train/sample_000031/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..40f89563efce1f3e1a4fa51fa46d2f6dfb0ade1c
--- /dev/null
+++ b/train/sample_000031/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4342a4c2339e8956b044d8dc99e1337301166e91e64ca367348d703c8ea95c59
+size 30623
diff --git a/train/sample_000031/board_t_plus_1.png b/train/sample_000031/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..6bb46fc2ead2e8449b47c248921816b6363f15e5
--- /dev/null
+++ b/train/sample_000031/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a1eda7889fd48d1a0fefe28247a15541b1ac97d460e1889d8427b6a1bac91e88
+size 30498
diff --git a/train/sample_000031/metadata.json b/train/sample_000031/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..20492475f2805304d7e5a40b99947e82878af967
--- /dev/null
+++ b/train/sample_000031/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000031",
+  "puzzle_id": "rvnIg",
+  "task": "task3",
+  "fen": "8/7p/2pk2p1/p2p1pP1/P2P4/2P1KPP1/8/8 b - - 0 48",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "c5",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/validation/sample_000000/board_t.png b/validation/sample_000000/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..dfd9a80c22440bc7fcb8781c6ce0f6e5fd6d44ce
--- /dev/null
+++ b/validation/sample_000000/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1f23249ec939f18070be1ed832491c6c4f824878ab3133381cf0dc8e42b7e7a1
+size 40206
diff --git a/validation/sample_000000/board_t_plus_1.png b/validation/sample_000000/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..b24af3e06ee0d1e13581d2370dc3fc39d9ab023f
--- /dev/null
+++ b/validation/sample_000000/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cd2bc8790aaa7503e43f7a9a1c46447a8ee7888ca3fd40c33c3a7010bbb04326
+size 39552
diff --git a/validation/sample_000000/metadata.json b/validation/sample_000000/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..86976c5540215533e5c01f0710176d50ef017e3c
--- /dev/null
+++ b/validation/sample_000000/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000000",
+  "puzzle_id": "9JiYC",
+  "task": "task3",
+  "fen": "1r2r1k1/5pp1/2b2q1p/4R3/8/2P1Q3/PP3PPP/R1B3K1 w - - 1 21",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "Rxe8+",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/validation/sample_000001/board_t.png b/validation/sample_000001/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..3c19518e3532155009363ba00466251039ac8e30
--- /dev/null
+++ b/validation/sample_000001/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b8dbfafef5fc6e223e00c051990e3c21a0da53f1c1157f795111865175af604b
+size 49996
diff --git a/validation/sample_000001/board_t_plus_1.png b/validation/sample_000001/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..469d39f695afad5dba9e71d09608373d27db9ded
--- /dev/null
+++ b/validation/sample_000001/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:521c0fc426036ec231be16cc9c9480556f0257d5b9c1c868327ce363328fd7ee
+size 50001
diff --git a/validation/sample_000001/metadata.json b/validation/sample_000001/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..10fe79b74c8bec06588d60f10c999aae8768b9d0
--- /dev/null
+++ b/validation/sample_000001/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000001",
+  "puzzle_id": "Lhld2",
+  "task": "task3",
+  "fen": "r4rk1/1bq1bppp/p4n2/1p2p1B1/4P3/1PN2N1P/1P2QPP1/2RR2K1 b - - 3 16",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "Rad8",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/validation/sample_000002/board_t.png b/validation/sample_000002/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..80c8f7781d3db3e135356bba75a56232610b0517
--- /dev/null
+++ b/validation/sample_000002/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aa018b3a34e9d12a5e3ad7edb967da0e335e3f8a723143bee28c665ab056b6e2
+size 43708
diff --git a/validation/sample_000002/board_t_plus_1.png b/validation/sample_000002/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..77c9f78a005440de9d6268d652cb3a059026525f
--- /dev/null
+++ b/validation/sample_000002/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c26b7328644d3b1f14a49fb760f13671a21f7ab6806f35e1397bc9fc5d547bea
+size 42400
diff --git a/validation/sample_000002/metadata.json b/validation/sample_000002/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..62d0f5d0ca25a3a891ff0dfb239244d9e37af7e2
--- /dev/null
+++ b/validation/sample_000002/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000002",
+  "puzzle_id": "RdgvN",
+  "task": "task3",
+  "fen": "r2b1r1k/p5pp/2Qp4/4pP2/2n5/2P3N1/PPq2P1P/3R1RK1 b - - 7 24",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "Nxb2",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file
diff --git a/validation/sample_000003/board_t.png b/validation/sample_000003/board_t.png
new file mode 100644
index 0000000000000000000000000000000000000000..9463782b15c322de90be9541b59fa4b0c31062c7
--- /dev/null
+++ b/validation/sample_000003/board_t.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f1bf8fce2a479b433f67cdc25f3d5d137bc341a3dc17ecced4a4fc84ac2ec9fd
+size 51820
diff --git a/validation/sample_000003/board_t_plus_1.png b/validation/sample_000003/board_t_plus_1.png
new file mode 100644
index 0000000000000000000000000000000000000000..e73652a9f1afd7efc48382ea017b70a82ab8079f
--- /dev/null
+++ b/validation/sample_000003/board_t_plus_1.png
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d32b567463de98e571584fa574ec1b4f615a8530b109fc0fdaaf8486ca14d301
+size 51588
diff --git a/validation/sample_000003/metadata.json b/validation/sample_000003/metadata.json
new file mode 100644
index 0000000000000000000000000000000000000000..75c83f5da4daa6dff6fcd400a36757ba4df555ca
--- /dev/null
+++ b/validation/sample_000003/metadata.json
@@ -0,0 +1,16 @@
+{
+  "sample_id": "sample_000003",
+  "puzzle_id": "YKK0p",
+  "task": "task3",
+  "fen": "rn3rk1/pb2ppbp/6p1/1ppqP1N1/3P4/4B3/PPP1B1PP/R2Q1RK1 w - - 0 13",
+  "prompt": "You are a specialized model for chessboard temporal reasoning.\nYour goal is to identify the move that transitioned the chessboard from State t to State t+1.\nInput:\n- Frame 1: Chessboard state at time t.\n- Frame 2: Chessboard state at time t+1.\nOutput Format:\nReturn only the move in Standard Algebraic Notation (SAN).",
+  "target": "Bd3",
+  "image_count": 2,
+  "image_files": [
+    "board_t.png",
+    "board_t_plus_1.png"
+  ],
+  "image_size": 512,
+  "patch_order": "row_major",
+  "patch_size": 16
+}
\ No newline at end of file