Training in progress, step 1000

config.json

@@ -7,6 +7,7 @@
   "attention_dropout": 0.0,
   "bos_token_id": 32013,
   "eos_token_id": 32021,
+  "head_dim": 128,
   "hidden_act": "silu",
   "hidden_size": 2048,
   "initializer_range": 0.02,
@@ -27,7 +28,7 @@
   "rope_theta": 100000,
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
-  "transformers_version": "4.44.2",
+  "transformers_version": "4.46.3",
   "use_cache": false,
   "vocab_size": 32092
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:304356823d58df9c3dd75e72387f95040c7e023a0482357738e8169117ad6f93
+oid sha256:125b8f48a404b64be11eee33e48d97bfcda1b389bccbe84fbf20654252d8dafe
 size 2691625640

tokenizer_config.json

@@ -180,7 +180,7 @@
       "special": true
     },
     "32022": {
-      "content": "ldr",
+      "content": "r15",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -188,7 +188,7 @@
       "special": false
     },
     "32023": {
-      "content": "c2",
+      "content": "bh",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -196,7 +196,7 @@
       "special": false
     },
     "32024": {
-      "content": "blx",
+      "content": "sar",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -204,7 +204,7 @@
       "special": false
     },
     "32025": {
-      "content": "jle",
+      "content": "ebx",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -212,7 +212,7 @@
       "special": false
     },
     "32026": {
-      "content": "r3",
+      "content": "ldr",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -220,7 +220,7 @@
       "special": false
     },
     "32027": {
-      "content": "r4",
+      "content": "r7",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -228,7 +228,7 @@
       "special": false
     },
     "32028": {
-      "content": "ebx",
+      "content": "r12d",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -236,7 +236,7 @@
       "special": false
     },
     "32029": {
-      "content": "rbp",
+      "content": "r1",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -244,7 +244,7 @@
       "special": false
     },
     "32030": {
-      "content": "bgt",
+      "content": "eax",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -252,7 +252,7 @@
       "special": false
     },
     "32031": {
-      "content": "r13",
+      "content": "c3",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -260,7 +260,7 @@
       "special": false
     },
     "32032": {
-      "content": "cmp",
+      "content": "r14d",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -268,7 +268,7 @@
       "special": false
     },
     "32033": {
-      "content": "xor",
+      "content": "r10",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -276,7 +276,7 @@
       "special": false
     },
     "32034": {
-      "content": "r8d",
+      "content": "rcx",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -284,7 +284,7 @@
       "special": false
     },
     "32035": {
-      "content": "r11",
+      "content": "rdi",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -292,7 +292,7 @@
       "special": false
     },
     "32036": {
-      "content": "nop",
+      "content": "rdx",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -300,7 +300,7 @@
       "special": false
     },
     "32037": {
-      "content": "sar",
+      "content": "bge",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -308,7 +308,7 @@
       "special": false
     },
     "32038": {
-      "content": "r0",
+      "content": "r8",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -316,7 +316,7 @@
       "special": false
     },
     "32039": {
-      "content": "rdx",
+      "content": "lr",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -324,7 +324,7 @@
       "special": false
     },
     "32040": {
-      "content": "edx",
+      "content": "r10d",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -332,7 +332,7 @@
       "special": false
     },
     "32041": {
-      "content": "r9",
+      "content": "jmp",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -340,7 +340,7 @@
       "special": false
     },
     "32042": {
-      "content": "r12",
+      "content": "beq",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -348,7 +348,7 @@
       "special": false
     },
     "32043": {
-      "content": "eax",
+      "content": "c5",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -356,7 +356,7 @@
       "special": false
     },
     "32044": {
-      "content": "c5",
+      "content": "r9d",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -364,7 +364,7 @@
       "special": false
     },
     "32045": {
-      "content": "r14",
+      "content": "r4",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -372,7 +372,7 @@
       "special": false
     },
     "32046": {
-      "content": "c4",
+      "content": "r11d",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -380,7 +380,7 @@
       "special": false
     },
     "32047": {
-      "content": "rsi",
+      "content": "blx",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -388,7 +388,7 @@
       "special": false
     },
     "32048": {
-      "content": "rsp",
+      "content": "r0",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -396,7 +396,7 @@
       "special": false
     },
     "32049": {
-      "content": "rbx",
+      "content": "jle",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -404,7 +404,7 @@
       "special": false
     },
     "32050": {
-      "content": "c3",
+      "content": "xor",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -412,7 +412,7 @@
       "special": false
     },
     "32051": {
-      "content": "r10d",
+      "content": "r12",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -420,7 +420,7 @@
       "special": false
     },
     "32052": {
-      "content": "jg",
+      "content": "r14",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -428,7 +428,7 @@
       "special": false
     },
     "32053": {
-      "content": "r8",
+      "content": "r15d",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -436,7 +436,7 @@
       "special": false
     },
     "32054": {
-      "content": "bne",
+      "content": "r6",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -444,7 +444,7 @@
       "special": false
     },
     "32055": {
-      "content": "r11d",
+      "content": "swi",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -452,7 +452,7 @@
       "special": false
     },
     "32056": {
-      "content": "QWORD",
+      "content": "r13",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -460,7 +460,7 @@
       "special": false
     },
     "32057": {
-      "content": "r7",
+      "content": "bgt",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -468,7 +468,7 @@
       "special": false
     },
     "32058": {
-      "content": "r13d",
+      "content": "jne",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -476,7 +476,7 @@
       "special": false
     },
     "32059": {
-      "content": "shr",
+      "content": "r11",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -484,7 +484,7 @@
       "special": false
     },
     "32060": {
-      "content": "c6",
+      "content": "c0",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -492,7 +492,7 @@
       "special": false
     },
     "32061": {
-      "content": "lr",
+      "content": "nop",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -508,7 +508,7 @@
       "special": false
     },
     "32063": {
-      "content": "cx",
+      "content": "c4",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -516,7 +516,7 @@
       "special": false
     },
     "32064": {
-      "content": "r5",
+      "content": "jg",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -524,7 +524,7 @@
       "special": false
     },
     "32065": {
-      "content": "r10",
+      "content": "DWORD",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -532,7 +532,7 @@
       "special": false
     },
     "32066": {
-      "content": "rcx",
+      "content": "r2",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -540,7 +540,7 @@
       "special": false
     },
     "32067": {
-      "content": "r15",
+      "content": "PTR",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -548,7 +548,7 @@
       "special": false
     },
     "32068": {
-      "content": "rdi",
+      "content": "jge",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -556,7 +556,7 @@
       "special": false
     },
     "32069": {
-      "content": "c0",
+      "content": "edx",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -564,7 +564,7 @@
       "special": false
     },
     "32070": {
-      "content": "swi",
+      "content": "cx",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -572,7 +572,7 @@
       "special": false
     },
     "32071": {
-      "content": "r2",
+      "content": "c2",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -580,7 +580,7 @@
       "special": false
     },
     "32072": {
-      "content": "r1",
+      "content": "shr",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -588,7 +588,7 @@
       "special": false
     },
     "32073": {
-      "content": "jmp",
+      "content": "edi",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -596,7 +596,7 @@
       "special": false
     },
     "32074": {
-      "content": "DWORD",
+      "content": "r3",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -604,7 +604,7 @@
       "special": false
     },
     "32075": {
-      "content": "PTR",
+      "content": "ecx",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -612,7 +612,7 @@
       "special": false
     },
     "32076": {
-      "content": "edi",
+      "content": "r8d",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -620,7 +620,7 @@
       "special": false
     },
     "32077": {
-      "content": "rax",
+      "content": "QWORD",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -628,7 +628,7 @@
       "special": false
     },
     "32078": {
-      "content": "bge",
+      "content": "rbx",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -636,7 +636,7 @@
       "special": false
     },
     "32079": {
-      "content": "jge",
+      "content": "mul",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -644,7 +644,7 @@
       "special": false
     },
     "32080": {
-      "content": "ecx",
+      "content": "rax",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -652,7 +652,7 @@
       "special": false
     },
     "32081": {
-      "content": "r9d",
+      "content": "rbp",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -660,7 +660,7 @@
       "special": false
     },
     "32082": {
-      "content": "r6",
+      "content": "bne",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -668,7 +668,7 @@
       "special": false
     },
     "32083": {
-      "content": "blt",
+      "content": "r5",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -676,7 +676,7 @@
       "special": false
     },
     "32084": {
-      "content": "jne",
+      "content": "r9",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -692,7 +692,7 @@
       "special": false
     },
     "32086": {
-      "content": "bh",
+      "content": "c6",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -700,7 +700,7 @@
       "special": false
     },
     "32087": {
-      "content": "r12d",
+      "content": "blt",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -708,7 +708,7 @@
       "special": false
     },
     "32088": {
-      "content": "r14d",
+      "content": "cmp",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -716,7 +716,7 @@
       "special": false
     },
     "32089": {
-      "content": "beq",
+      "content": "r13d",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -724,7 +724,7 @@
       "special": false
     },
     "32090": {
-      "content": "mul",
+      "content": "rsp",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -732,7 +732,7 @@
       "special": false
     },
     "32091": {
-      "content": "r15d",
+      "content": "rsi",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:71bdea7155c86413006235525141312ddd2048baf0eb878c33c7d06ee0f56e34
-size 5432
+oid sha256:ce299981a75e3ad3bcafb97ab6b44f22b8a06b567416185cc97b2647853ff879
+size 5560