Upload LlamaForCausalLM
Browse files- config.json +8 -1
config.json
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
{
|
| 2 |
-
"_name_or_path": "/home/ubuntu/onebitllms/ckpts-to-push/3b-instruct
|
| 3 |
"architectures": [
|
| 4 |
"LlamaForCausalLM"
|
| 5 |
],
|
|
@@ -12,6 +12,7 @@
|
|
| 12 |
"hidden_size": 2048,
|
| 13 |
"initializer_range": 0.02,
|
| 14 |
"intermediate_size": 13312,
|
|
|
|
| 15 |
"max_position_embeddings": 32768,
|
| 16 |
"mlp_bias": false,
|
| 17 |
"model_type": "llama",
|
|
@@ -19,6 +20,12 @@
|
|
| 19 |
"num_hidden_layers": 32,
|
| 20 |
"num_key_value_heads": 2,
|
| 21 |
"pretraining_tp": 1,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 22 |
"rms_norm_eps": 1e-05,
|
| 23 |
"rope_scaling": null,
|
| 24 |
"rope_theta": 1000000,
|
|
|
|
| 1 |
{
|
| 2 |
+
"_name_or_path": "/home/ubuntu/onebitllms/ckpts-to-push/3b-instruct",
|
| 3 |
"architectures": [
|
| 4 |
"LlamaForCausalLM"
|
| 5 |
],
|
|
|
|
| 12 |
"hidden_size": 2048,
|
| 13 |
"initializer_range": 0.02,
|
| 14 |
"intermediate_size": 13312,
|
| 15 |
+
"is_bitnet_config": true,
|
| 16 |
"max_position_embeddings": 32768,
|
| 17 |
"mlp_bias": false,
|
| 18 |
"model_type": "llama",
|
|
|
|
| 20 |
"num_hidden_layers": 32,
|
| 21 |
"num_key_value_heads": 2,
|
| 22 |
"pretraining_tp": 1,
|
| 23 |
+
"quantization_config": {
|
| 24 |
+
"modules_to_not_convert": [
|
| 25 |
+
"lm_head"
|
| 26 |
+
],
|
| 27 |
+
"quant_method": "bitnet"
|
| 28 |
+
},
|
| 29 |
"rms_norm_eps": 1e-05,
|
| 30 |
"rope_scaling": null,
|
| 31 |
"rope_theta": 1000000,
|