Update config.json
Browse files- config.json +12 -17
config.json
CHANGED
|
@@ -1,24 +1,13 @@
|
|
| 1 |
{
|
| 2 |
-
"model_type": "
|
| 3 |
-
"architectures": ["GPT2LMHeadModel"],
|
| 4 |
"vocab_size": 1920,
|
| 5 |
-
"
|
|
|
|
|
|
|
|
|
|
| 6 |
"n_ctx": 64,
|
| 7 |
-
"
|
| 8 |
-
"n_layer": 6,
|
| 9 |
-
"n_head": 6,
|
| 10 |
-
"activation_function": "gelu_new",
|
| 11 |
-
"resid_pdrop": 0.0,
|
| 12 |
-
"embd_pdrop": 0.0,
|
| 13 |
-
"attn_pdrop": 0.0,
|
| 14 |
-
"layer_norm_epsilon": 1e-5,
|
| 15 |
-
"initializer_range": 0.02,
|
| 16 |
-
"use_cache": true,
|
| 17 |
-
"bos_token_id": 1,
|
| 18 |
-
"eos_token_id": 2,
|
| 19 |
-
"pad_token_id": 0,
|
| 20 |
"tie_word_embeddings": true,
|
| 21 |
-
"transformers_version": "4.45.0",
|
| 22 |
"special_tokens": {
|
| 23 |
"pad_token": "<PAD>",
|
| 24 |
"bos_token": "<BOS>",
|
|
@@ -32,5 +21,11 @@
|
|
| 32 |
"sep": 3,
|
| 33 |
"unk": 4
|
| 34 |
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 35 |
}
|
| 36 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"model_type": "tiny-gpt",
|
|
|
|
| 3 |
"vocab_size": 1920,
|
| 4 |
+
"n_layers": 6,
|
| 5 |
+
"n_heads": 6,
|
| 6 |
+
"d_model": 240,
|
| 7 |
+
"d_ff": 960,
|
| 8 |
"n_ctx": 64,
|
| 9 |
+
"dropout": 0.0,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 10 |
"tie_word_embeddings": true,
|
|
|
|
| 11 |
"special_tokens": {
|
| 12 |
"pad_token": "<PAD>",
|
| 13 |
"bos_token": "<BOS>",
|
|
|
|
| 21 |
"sep": 3,
|
| 22 |
"unk": 4
|
| 23 |
}
|
| 24 |
+
},
|
| 25 |
+
"quantization": {
|
| 26 |
+
"default": "int8",
|
| 27 |
+
"embedding_bits": 18,
|
| 28 |
+
"layernorm_bits": 24,
|
| 29 |
+
"format": "safetensors+scales"
|
| 30 |
}
|
| 31 |
}
|