Update config.json
Browse filesUpdated configuration for SequenceClassification
- config.json +2 -1
config.json
CHANGED
|
@@ -1,7 +1,7 @@
|
|
| 1 |
{
|
| 2 |
"_name_or_path": "/data1/rsaparkh/saved_models/llama3-8b/",
|
| 3 |
"architectures": [
|
| 4 |
-
"
|
| 5 |
],
|
| 6 |
"attention_bias": false,
|
| 7 |
"attention_dropout": 0.0,
|
|
@@ -27,6 +27,7 @@
|
|
| 27 |
"num_attention_heads": 32,
|
| 28 |
"num_hidden_layers": 32,
|
| 29 |
"num_key_value_heads": 8,
|
|
|
|
| 30 |
"pretraining_tp": 1,
|
| 31 |
"quantization_config": {
|
| 32 |
"_load_in_4bit": false,
|
|
|
|
| 1 |
{
|
| 2 |
"_name_or_path": "/data1/rsaparkh/saved_models/llama3-8b/",
|
| 3 |
"architectures": [
|
| 4 |
+
"LlamaForSequenceClassification"
|
| 5 |
],
|
| 6 |
"attention_bias": false,
|
| 7 |
"attention_dropout": 0.0,
|
|
|
|
| 27 |
"num_attention_heads": 32,
|
| 28 |
"num_hidden_layers": 32,
|
| 29 |
"num_key_value_heads": 8,
|
| 30 |
+
"num_labels": 3, // Added number of labels for classification
|
| 31 |
"pretraining_tp": 1,
|
| 32 |
"quantization_config": {
|
| 33 |
"_load_in_4bit": false,
|