Update config.json
Browse files- config.json +48 -48
config.json
CHANGED
@@ -1,48 +1,48 @@
|
|
1 |
-
{
|
2 |
-
"_name_or_path": "
|
3 |
-
"architectures": [
|
4 |
-
"MolmoForCausalLM"
|
5 |
-
],
|
6 |
-
"attention_layer_norm": false,
|
7 |
-
"auto_map": {
|
8 |
-
"AutoConfig": "config_molmo.MolmoConfig",
|
9 |
-
"AutoModelForCausalLM": "
|
10 |
-
},
|
11 |
-
"clip_qkv": null,
|
12 |
-
"embedding_size": 152064,
|
13 |
-
"hidden_size": 3584,
|
14 |
-
"initializer_range": 0.02,
|
15 |
-
"intermediate_size": 37888,
|
16 |
-
"layer_norm_eps": 1e-06,
|
17 |
-
"layer_norm_type": "rms",
|
18 |
-
"max_position_embeddings": 4096,
|
19 |
-
"model_type": "molmo",
|
20 |
-
"norm_after": false,
|
21 |
-
"num_attention_heads": 28,
|
22 |
-
"num_hidden_layers": 28,
|
23 |
-
"num_key_value_heads": 4,
|
24 |
-
"qkv_bias": true,
|
25 |
-
"quantization_config": {
|
26 |
-
"_load_in_4bit": true,
|
27 |
-
"_load_in_8bit": false,
|
28 |
-
"bnb_4bit_compute_dtype": "float32",
|
29 |
-
"bnb_4bit_quant_storage": "uint8",
|
30 |
-
"bnb_4bit_quant_type": "fp4",
|
31 |
-
"bnb_4bit_use_double_quant": true,
|
32 |
-
"llm_int8_enable_fp32_cpu_offload": false,
|
33 |
-
"llm_int8_has_fp16_weight": false,
|
34 |
-
"llm_int8_skip_modules": null,
|
35 |
-
"llm_int8_threshold": 6.0,
|
36 |
-
"load_in_4bit": true,
|
37 |
-
"load_in_8bit": false,
|
38 |
-
"quant_method": "bitsandbytes"
|
39 |
-
},
|
40 |
-
"rope_theta": 1000000.0,
|
41 |
-
"tie_word_embeddings": false,
|
42 |
-
"torch_dtype": "float32",
|
43 |
-
"transformers_version": "4.44.0",
|
44 |
-
"use_cache": true,
|
45 |
-
"use_position_ids": true,
|
46 |
-
"vocab_size": 152064,
|
47 |
-
"weight_tying": false
|
48 |
-
}
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "cyan2k/molmo-7B-D-bnb-4bit",
|
3 |
+
"architectures": [
|
4 |
+
"MolmoForCausalLM"
|
5 |
+
],
|
6 |
+
"attention_layer_norm": false,
|
7 |
+
"auto_map": {
|
8 |
+
"AutoConfig": "config_molmo.MolmoConfig",
|
9 |
+
"AutoModelForCausalLM": "modeling_molmo.MolmoForCausalLM"
|
10 |
+
},
|
11 |
+
"clip_qkv": null,
|
12 |
+
"embedding_size": 152064,
|
13 |
+
"hidden_size": 3584,
|
14 |
+
"initializer_range": 0.02,
|
15 |
+
"intermediate_size": 37888,
|
16 |
+
"layer_norm_eps": 1e-06,
|
17 |
+
"layer_norm_type": "rms",
|
18 |
+
"max_position_embeddings": 4096,
|
19 |
+
"model_type": "molmo",
|
20 |
+
"norm_after": false,
|
21 |
+
"num_attention_heads": 28,
|
22 |
+
"num_hidden_layers": 28,
|
23 |
+
"num_key_value_heads": 4,
|
24 |
+
"qkv_bias": true,
|
25 |
+
"quantization_config": {
|
26 |
+
"_load_in_4bit": true,
|
27 |
+
"_load_in_8bit": false,
|
28 |
+
"bnb_4bit_compute_dtype": "float32",
|
29 |
+
"bnb_4bit_quant_storage": "uint8",
|
30 |
+
"bnb_4bit_quant_type": "fp4",
|
31 |
+
"bnb_4bit_use_double_quant": true,
|
32 |
+
"llm_int8_enable_fp32_cpu_offload": false,
|
33 |
+
"llm_int8_has_fp16_weight": false,
|
34 |
+
"llm_int8_skip_modules": null,
|
35 |
+
"llm_int8_threshold": 6.0,
|
36 |
+
"load_in_4bit": true,
|
37 |
+
"load_in_8bit": false,
|
38 |
+
"quant_method": "bitsandbytes"
|
39 |
+
},
|
40 |
+
"rope_theta": 1000000.0,
|
41 |
+
"tie_word_embeddings": false,
|
42 |
+
"torch_dtype": "float32",
|
43 |
+
"transformers_version": "4.44.0",
|
44 |
+
"use_cache": true,
|
45 |
+
"use_position_ids": true,
|
46 |
+
"vocab_size": 152064,
|
47 |
+
"weight_tying": false
|
48 |
+
}
|