Upload folder using huggingface_hub
Browse files- README.md +2 -2
- hyperparameters.json +1 -0
- tokenizer.json +28 -2
README.md
CHANGED
|
@@ -7,12 +7,12 @@ library_name: transformers
|
|
| 7 |
Chain of Thought (CoT) transformer model trained to do multi-step integer arithmetic.
|
| 8 |
|
| 9 |
Model details:
|
| 10 |
-
- **Vocabulary Size**:
|
| 11 |
- **Layer Count**: 8
|
| 12 |
- **Attention Head Count**: 4
|
| 13 |
- **Residual Stream Size**: 256
|
| 14 |
- **Context Length**: 256
|
| 15 |
-
- **Tokens Trained on**: 419,
|
| 16 |
|
| 17 |
Training Score During Training
|
| 18 |
|
|
|
|
| 7 |
Chain of Thought (CoT) transformer model trained to do multi-step integer arithmetic.
|
| 8 |
|
| 9 |
Model details:
|
| 10 |
+
- **Vocabulary Size**: 40 (Character Tokenization)
|
| 11 |
- **Layer Count**: 8
|
| 12 |
- **Attention Head Count**: 4
|
| 13 |
- **Residual Stream Size**: 256
|
| 14 |
- **Context Length**: 256
|
| 15 |
+
- **Tokens Trained on**: 419,612,160
|
| 16 |
|
| 17 |
Training Score During Training
|
| 18 |
|
hyperparameters.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"MIN_DIFFICULTY": 2, "MAX_DIFFICULTY": 4, "TRAINING_SAMPLES": 3000000, "CONTEXT_LENGTH": 256, "RESIDUAL_EMBEDDING_SIZE": 256, "MLP_EMBEDDING_SIZE": 1024, "NUM_ATTENTION_HEADS": 4, "NUM_LAYERS": 8, "VOCAB_SIZE": 40, "TOTAL_TOKENS": 419612160}
|
tokenizer.json
CHANGED
|
@@ -112,8 +112,34 @@
|
|
| 112 |
"t": 36,
|
| 113 |
"u": 37,
|
| 114 |
"Ċ": 38,
|
| 115 |
-
"Ġ": 39
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 116 |
},
|
| 117 |
-
"merges": [
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 118 |
}
|
| 119 |
}
|
|
|
|
| 112 |
"t": 36,
|
| 113 |
"u": 37,
|
| 114 |
"Ċ": 38,
|
| 115 |
+
"Ġ": 39,
|
| 116 |
+
"Ġ-": 40,
|
| 117 |
+
"(-": 41,
|
| 118 |
+
"Ġ1": 42,
|
| 119 |
+
"St": 43,
|
| 120 |
+
"ep": 44
|
| 121 |
},
|
| 122 |
+
"merges": [
|
| 123 |
+
[
|
| 124 |
+
"Ġ",
|
| 125 |
+
"-"
|
| 126 |
+
],
|
| 127 |
+
[
|
| 128 |
+
"(",
|
| 129 |
+
"-"
|
| 130 |
+
],
|
| 131 |
+
[
|
| 132 |
+
"Ġ",
|
| 133 |
+
"1"
|
| 134 |
+
],
|
| 135 |
+
[
|
| 136 |
+
"S",
|
| 137 |
+
"t"
|
| 138 |
+
],
|
| 139 |
+
[
|
| 140 |
+
"e",
|
| 141 |
+
"p"
|
| 142 |
+
]
|
| 143 |
+
]
|
| 144 |
}
|
| 145 |
}
|