Mini.Deep.Thinker.11m | 20,000 new examples | step 1,250

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+seen_examples.jsonl filter=lfs diff=lfs merge=lfs -text

README.md

@@ -63,9 +63,9 @@ Training state is persisted to Hugging Face, including:
 
 ## Current progress
 
-- Unique examples trained: 20,000
-- Global optimizer steps: 625
+- Unique examples trained: 40,000
+- Global optimizer steps: 1,250
 - Last session: 20,000 examples
-- Best session loss: 3.06134
+- Best session loss: 1.88992
 
 This is an experimental small language model and is not guaranteed to produce factually or logically correct outputs.

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7bc26d2dd56ce812cd2ea3f039fa6bd5e2e476d09779360113b7dc58ac353267
+oid sha256:1be18bc9af10fd3159e514d02469f6b5521f1176281be6130bf642829610e5eb
 size 49794720

progress.json

@@ -5,36 +5,36 @@
   "context_length": 1096,
   "vocab_size": 4096,
   "parameter_count": 11004896,
-  "global_step": 625,
-  "unique_examples_completed": 20000,
+  "global_step": 1250,
+  "unique_examples_completed": 40000,
   "last_session_examples": 20000,
   "max_session_examples": 20000,
   "last_session_category_distribution": {
-    "Instruct": 4952,
-    "Code_Instruct": 12819,
-    "Think": 1486,
-    "Reasoning": 165,
-    "Chat": 332,
-    "Thought": 205,
-    "Tool_Calling": 41
+    "Code_Instruct": 12741,
+    "Instruct": 4926,
+    "Thought": 179,
+    "Think": 1607,
+    "Chat": 326,
+    "Reasoning": 176,
+    "Tool_Calling": 45
   },
   "last_session_source_distribution": {
-    "Plans11/Organized_PreTrain_WIUAI_1.3M": 10290,
-    "Plans11/Organized_PreTrain_Cyber_Security_640k": 5077,
-    "Plans11/Organized_PreTrain_Instruct_366k": 2809,
-    "Plans11/Organized_PreTrain_Frontier_Traces_47k": 311,
-    "Plans11/Organized_PreTrain_Frontier_Traces_150k": 494,
-    "Plans11/Organized_Pretrain_HR_213K": 254,
-    "Plans11/Organized_PreTrain_Frontier_Traces_34K": 119,
-    "Plans11/Organized_PreTrain_Persona_GOD_Seed_52K": 413,
-    "Plans11/Organized_PreTrain_Frontier_Traces_33k": 52,
-    "Plans11/Organized_PreTrain_Agent_287k": 41,
-    "Plans11/Organized_PreTrain_Coding_239k": 140
+    "Plans11/Organized_PreTrain_WIUAI_1.3M": 10364,
+    "Plans11/Organized_PreTrain_Cyber_Security_640k": 4991,
+    "Plans11/Organized_PreTrain_Instruct_366k": 2789,
+    "Plans11/Organized_PreTrain_Frontier_Traces_150k": 523,
+    "Plans11/Organized_Pretrain_HR_213K": 243,
+    "Plans11/Organized_PreTrain_Frontier_Traces_47k": 341,
+    "Plans11/Organized_PreTrain_Persona_GOD_Seed_52K": 407,
+    "Plans11/Organized_PreTrain_Frontier_Traces_34K": 98,
+    "Plans11/Organized_PreTrain_Coding_239k": 137,
+    "Plans11/Organized_PreTrain_Agent_287k": 45,
+    "Plans11/Organized_PreTrain_Frontier_Traces_33k": 62
   },
-  "timestamp": 1787146729.219449,
+  "timestamp": 1787150298.2830229,
   "causal_next_token": true,
   "strict_example_context": true,
   "tokenizer_immutable": true,
   "shuffled_deterministic_scan": true,
-  "best_session_loss": 3.061336768269539
+  "best_session_loss": 1.889919606000185
 }

training_metadata.json

@@ -6,8 +6,8 @@
   "strict_example_limit": 1095,
   "vocab_size": 4096,
   "dataset": "Plans11/Organized_PreTrain_1k_Context",
-  "global_step": 625,
-  "unique_examples_trained": 20000,
+  "global_step": 1250,
+  "unique_examples_trained": 40000,
   "last_session_size": 20000,
   "learning_rate": 0.0003,
   "batch_size": 8,
@@ -24,5 +24,5 @@
   "tokenizer_immutable": true,
   "content_hash_deduplication": true,
   "shuffled_deterministic_scan": true,
-  "best_session_loss": 3.061336768269539
+  "best_session_loss": 1.889919606000185
 }

training_state.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0fc392f4b98a53da2d7d1d54e0406f7b303e0f825e90604150d6de6d104a553f
-size 132168635
+oid sha256:066eea6aa8152a4365b78c54d9e3f2ddc025f7acdea6b2b93a19b6da17a1eb66
+size 132169595