Hugging Face's logo

Fraser
/
LLaDA-8B-Recursive-ARC

Text Generation
Transformers
Safetensors
recursive-mlm
feature-extraction
conversational
custom_code
Model card Files
xet
 Community
Fraser commited on
Commit
67d3b9f
·
verified ·
1 Parent(s): 395bc45

Upload folder using huggingface_hub

Browse files
Files changed (6) hide show
  1. chat_template.jinja +11 -0
  2. config.json +153 -0
  3. model.safetensors.index.json +751 -0
  4. special_tokens_map.json +46 -0
  5. tokenizer.json +217 -0
  6. tokenizer_config.json +113 -0
chat_template.jinja ADDED
@@ -0,0 +1,11 @@
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {% set loop_messages = messages %}
2
+ {% for message in loop_messages %}
3
+ {% if loop.index0 == 0 %}{{ bos_token }}{% endif %}
4
+ <|start_header_id|>{{ message['role'] }}<|end_header_id|>
5
+
6
+ {{ message['content'] | trim }}<|eot_id|>
7
+ {%- endfor %}
8
+ {% if add_generation_prompt and (loop_messages | length == 0 or loop_messages[-1]['role'] != 'assistant') %}
9
+ <|start_header_id|>assistant<|end_header_id|>
10
+
11
+ {% endif %}
config.json ADDED
@@ -0,0 +1,153 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "RecursiveMaskedLM"
4
+ ],
5
+ "base_model_config": {
6
+ "_name_or_path": "Fraser/LLaDA-8B-Base-gg2m",
7
+ "activation_type": "silu",
8
+ "add_cross_attention": false,
9
+ "alibi": false,
10
+ "alibi_bias_max": 8.0,
11
+ "architectures": [
12
+ "LLaDAModelLM"
13
+ ],
14
+ "attention_dropout": 0.0,
15
+ "attention_layer_norm": false,
16
+ "attention_layer_norm_with_affine": true,
17
+ "auto_map": {
18
+ "AutoConfig": "configuration_llada.LLaDAConfig",
19
+ "AutoModel": "modeling_llada.LLaDAModelLM",
20
+ "AutoModelForCausalLM": "modeling_llada.LLaDAModelLM"
21
+ },
22
+ "bad_words_ids": null,
23
+ "begin_suppress_tokens": null,
24
+ "bias_for_layer_norm": false,
25
+ "block_group_size": 1,
26
+ "block_type": "llama",
27
+ "bos_token_id": 75,
28
+ "chunk_size_feed_forward": 0,
29
+ "cross_attention_hidden_size": null,
30
+ "d_model": 4096,
31
+ "decoder_start_token_id": null,
32
+ "diversity_penalty": 0.0,
33
+ "do_sample": false,
34
+ "dtype": "bfloat16",
35
+ "early_stopping": false,
36
+ "embedding_dropout": 0.0,
37
+ "embedding_size": 126464,
38
+ "encoder_no_repeat_ngram_size": 0,
39
+ "eos_token_id": 76,
40
+ "exponential_decay_length_penalty": null,
41
+ "finetuning_task": null,
42
+ "flash_attention": false,
43
+ "forced_bos_token_id": null,
44
+ "forced_eos_token_id": null,
45
+ "id2label": {
46
+ "0": "LABEL_0",
47
+ "1": "LABEL_1"
48
+ },
49
+ "include_bias": false,
50
+ "include_qkv_bias": false,
51
+ "init_cutoff_factor": null,
52
+ "init_device": "meta",
53
+ "init_fn": "mitchell",
54
+ "init_std": 0.02,
55
+ "input_emb_norm": false,
56
+ "is_decoder": false,
57
+ "is_encoder_decoder": false,
58
+ "label2id": {
59
+ "LABEL_0": 0,
60
+ "LABEL_1": 1
61
+ },
62
+ "layer_norm_type": "rms",
63
+ "layer_norm_with_affine": true,
64
+ "length_penalty": 1.0,
65
+ "mask_token_id": 78,
66
+ "max_length": 20,
67
+ "max_sequence_length": 4096,
68
+ "min_length": 0,
69
+ "mlp_hidden_size": 12288,
70
+ "mlp_ratio": 4,
71
+ "model_type": "llada",
72
+ "multi_query_attention": null,
73
+ "n_heads": 32,
74
+ "n_kv_heads": 32,
75
+ "n_layers": 32,
76
+ "no_repeat_ngram_size": 0,
77
+ "num_beam_groups": 1,
78
+ "num_beams": 1,
79
+ "num_return_sequences": 1,
80
+ "output_attentions": false,
81
+ "output_hidden_states": false,
82
+ "output_scores": false,
83
+ "pad_token_id": 76,
84
+ "precision": "amp_bf16",
85
+ "prefix": null,
86
+ "problem_type": null,
87
+ "pruned_heads": {},
88
+ "remove_invalid_values": false,
89
+ "repetition_penalty": 1.0,
90
+ "residual_dropout": 0.0,
91
+ "return_dict": true,
92
+ "return_dict_in_generate": false,
93
+ "rms_norm_eps": 1e-05,
94
+ "rope": true,
95
+ "rope_full_precision": true,
96
+ "rope_theta": 500000.0,
97
+ "scale_logits": false,
98
+ "sep_token_id": null,
99
+ "suppress_tokens": null,
100
+ "task_specific_params": null,
101
+ "temperature": 1.0,
102
+ "tf_legacy_loss": false,
103
+ "tie_encoder_decoder": false,
104
+ "tie_word_embeddings": true,
105
+ "tokenizer_class": null,
106
+ "top_k": 50,
107
+ "top_p": 1.0,
108
+ "torchscript": false,
109
+ "transformers_version": "4.57.0",
110
+ "typical_p": 1.0,
111
+ "use_bfloat16": false,
112
+ "use_cache": false,
113
+ "vocab_size": 85,
114
+ "weight_tying": false
115
+ },
116
+ "bos_token_id": 75,
117
+ "causal_strength": 1.0,
118
+ "dtype": "bfloat16",
119
+ "entropy_floor_max": 0.0,
120
+ "entropy_target_max": 0.0,
121
+ "eos_token_id": 76,
122
+ "flow_matching_enabled": false,
123
+ "flow_matching_lambda": 0.5,
124
+ "flow_matching_mask_scale": false,
125
+ "flow_matching_noise_scale": 2.0,
126
+ "flow_matching_t_distribution": "logit_normal",
127
+ "flow_matching_t_logit_mean": -0.4,
128
+ "flow_matching_t_logit_std": 1.0,
129
+ "flow_matching_t_max": 0.99,
130
+ "flow_matching_t_min": 0.01,
131
+ "gradient_steps": null,
132
+ "iteration_rope_dim_fraction": 0.0,
133
+ "loss_weight": "linear",
134
+ "mask_token_id": 78,
135
+ "model_type": "recursive-mlm",
136
+ "noise_std_max": 0.0,
137
+ "normalization": "softmax",
138
+ "num_recursions": 4,
139
+ "pad_token_id": 76,
140
+ "schedule": "linear",
141
+ "self_distillation_enabled": false,
142
+ "self_distillation_lambda": 0.5,
143
+ "self_distillation_teacher": "first",
144
+ "self_distillation_temperature_distribution": "log_uniform",
145
+ "self_distillation_temperature_max": 10.0,
146
+ "self_distillation_temperature_min": 1.5,
147
+ "smear_sigma_max": 0.0,
148
+ "soft_embedding_ema_step": 1.0,
149
+ "soft_embedding_method": "softmax",
150
+ "temperature_max": 0.0,
151
+ "transformers_version": "4.57.0",
152
+ "use_recursion_checkpointing": true
153
+ }
model.safetensors.index.json ADDED
@@ -0,0 +1,751 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_parameters": 8326743040,
4
+ "total_size": 16653486080
5
+ },
6
+ "weight_map": {
7
+ "mlm.base_model.model.model.transformer.blocks.0.attn_norm.weight": "model-00001-of-00004.safetensors",
8
+ "mlm.base_model.model.model.transformer.blocks.0.attn_out.base_layer.weight": "model-00001-of-00004.safetensors",
9
+ "mlm.base_model.model.model.transformer.blocks.0.attn_out.lora_A.default.weight": "model-00001-of-00004.safetensors",
10
+ "mlm.base_model.model.model.transformer.blocks.0.attn_out.lora_B.default.weight": "model-00001-of-00004.safetensors",
11
+ "mlm.base_model.model.model.transformer.blocks.0.ff_norm.weight": "model-00001-of-00004.safetensors",
12
+ "mlm.base_model.model.model.transformer.blocks.0.ff_out.base_layer.weight": "model-00001-of-00004.safetensors",
13
+ "mlm.base_model.model.model.transformer.blocks.0.ff_out.lora_A.default.weight": "model-00001-of-00004.safetensors",
14
+ "mlm.base_model.model.model.transformer.blocks.0.ff_out.lora_B.default.weight": "model-00001-of-00004.safetensors",
15
+ "mlm.base_model.model.model.transformer.blocks.0.ff_proj.base_layer.weight": "model-00001-of-00004.safetensors",
16
+ "mlm.base_model.model.model.transformer.blocks.0.ff_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
17
+ "mlm.base_model.model.model.transformer.blocks.0.ff_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
18
+ "mlm.base_model.model.model.transformer.blocks.0.k_proj.base_layer.weight": "model-00001-of-00004.safetensors",
19
+ "mlm.base_model.model.model.transformer.blocks.0.k_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
20
+ "mlm.base_model.model.model.transformer.blocks.0.k_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
21
+ "mlm.base_model.model.model.transformer.blocks.0.q_proj.base_layer.weight": "model-00001-of-00004.safetensors",
22
+ "mlm.base_model.model.model.transformer.blocks.0.q_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
23
+ "mlm.base_model.model.model.transformer.blocks.0.q_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
24
+ "mlm.base_model.model.model.transformer.blocks.0.up_proj.base_layer.weight": "model-00001-of-00004.safetensors",
25
+ "mlm.base_model.model.model.transformer.blocks.0.up_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
26
+ "mlm.base_model.model.model.transformer.blocks.0.up_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
27
+ "mlm.base_model.model.model.transformer.blocks.0.v_proj.base_layer.weight": "model-00001-of-00004.safetensors",
28
+ "mlm.base_model.model.model.transformer.blocks.0.v_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
29
+ "mlm.base_model.model.model.transformer.blocks.0.v_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
30
+ "mlm.base_model.model.model.transformer.blocks.1.attn_norm.weight": "model-00001-of-00004.safetensors",
31
+ "mlm.base_model.model.model.transformer.blocks.1.attn_out.base_layer.weight": "model-00001-of-00004.safetensors",
32
+ "mlm.base_model.model.model.transformer.blocks.1.attn_out.lora_A.default.weight": "model-00001-of-00004.safetensors",
33
+ "mlm.base_model.model.model.transformer.blocks.1.attn_out.lora_B.default.weight": "model-00001-of-00004.safetensors",
34
+ "mlm.base_model.model.model.transformer.blocks.1.ff_norm.weight": "model-00001-of-00004.safetensors",
35
+ "mlm.base_model.model.model.transformer.blocks.1.ff_out.base_layer.weight": "model-00001-of-00004.safetensors",
36
+ "mlm.base_model.model.model.transformer.blocks.1.ff_out.lora_A.default.weight": "model-00001-of-00004.safetensors",
37
+ "mlm.base_model.model.model.transformer.blocks.1.ff_out.lora_B.default.weight": "model-00001-of-00004.safetensors",
38
+ "mlm.base_model.model.model.transformer.blocks.1.ff_proj.base_layer.weight": "model-00001-of-00004.safetensors",
39
+ "mlm.base_model.model.model.transformer.blocks.1.ff_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
40
+ "mlm.base_model.model.model.transformer.blocks.1.ff_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
41
+ "mlm.base_model.model.model.transformer.blocks.1.k_proj.base_layer.weight": "model-00001-of-00004.safetensors",
42
+ "mlm.base_model.model.model.transformer.blocks.1.k_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
43
+ "mlm.base_model.model.model.transformer.blocks.1.k_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
44
+ "mlm.base_model.model.model.transformer.blocks.1.q_proj.base_layer.weight": "model-00001-of-00004.safetensors",
45
+ "mlm.base_model.model.model.transformer.blocks.1.q_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
46
+ "mlm.base_model.model.model.transformer.blocks.1.q_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
47
+ "mlm.base_model.model.model.transformer.blocks.1.up_proj.base_layer.weight": "model-00001-of-00004.safetensors",
48
+ "mlm.base_model.model.model.transformer.blocks.1.up_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
49
+ "mlm.base_model.model.model.transformer.blocks.1.up_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
50
+ "mlm.base_model.model.model.transformer.blocks.1.v_proj.base_layer.weight": "model-00001-of-00004.safetensors",
51
+ "mlm.base_model.model.model.transformer.blocks.1.v_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
52
+ "mlm.base_model.model.model.transformer.blocks.1.v_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
53
+ "mlm.base_model.model.model.transformer.blocks.10.attn_norm.weight": "model-00002-of-00004.safetensors",
54
+ "mlm.base_model.model.model.transformer.blocks.10.attn_out.base_layer.weight": "model-00002-of-00004.safetensors",
55
+ "mlm.base_model.model.model.transformer.blocks.10.attn_out.lora_A.default.weight": "model-00002-of-00004.safetensors",
56
+ "mlm.base_model.model.model.transformer.blocks.10.attn_out.lora_B.default.weight": "model-00002-of-00004.safetensors",
57
+ "mlm.base_model.model.model.transformer.blocks.10.ff_norm.weight": "model-00002-of-00004.safetensors",
58
+ "mlm.base_model.model.model.transformer.blocks.10.ff_out.base_layer.weight": "model-00002-of-00004.safetensors",
59
+ "mlm.base_model.model.model.transformer.blocks.10.ff_out.lora_A.default.weight": "model-00002-of-00004.safetensors",
60
+ "mlm.base_model.model.model.transformer.blocks.10.ff_out.lora_B.default.weight": "model-00002-of-00004.safetensors",
61
+ "mlm.base_model.model.model.transformer.blocks.10.ff_proj.base_layer.weight": "model-00002-of-00004.safetensors",
62
+ "mlm.base_model.model.model.transformer.blocks.10.ff_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
63
+ "mlm.base_model.model.model.transformer.blocks.10.ff_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
64
+ "mlm.base_model.model.model.transformer.blocks.10.k_proj.base_layer.weight": "model-00002-of-00004.safetensors",
65
+ "mlm.base_model.model.model.transformer.blocks.10.k_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
66
+ "mlm.base_model.model.model.transformer.blocks.10.k_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
67
+ "mlm.base_model.model.model.transformer.blocks.10.q_proj.base_layer.weight": "model-00002-of-00004.safetensors",
68
+ "mlm.base_model.model.model.transformer.blocks.10.q_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
69
+ "mlm.base_model.model.model.transformer.blocks.10.q_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
70
+ "mlm.base_model.model.model.transformer.blocks.10.up_proj.base_layer.weight": "model-00002-of-00004.safetensors",
71
+ "mlm.base_model.model.model.transformer.blocks.10.up_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
72
+ "mlm.base_model.model.model.transformer.blocks.10.up_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
73
+ "mlm.base_model.model.model.transformer.blocks.10.v_proj.base_layer.weight": "model-00002-of-00004.safetensors",
74
+ "mlm.base_model.model.model.transformer.blocks.10.v_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
75
+ "mlm.base_model.model.model.transformer.blocks.10.v_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
76
+ "mlm.base_model.model.model.transformer.blocks.11.attn_norm.weight": "model-00002-of-00004.safetensors",
77
+ "mlm.base_model.model.model.transformer.blocks.11.attn_out.base_layer.weight": "model-00002-of-00004.safetensors",
78
+ "mlm.base_model.model.model.transformer.blocks.11.attn_out.lora_A.default.weight": "model-00002-of-00004.safetensors",
79
+ "mlm.base_model.model.model.transformer.blocks.11.attn_out.lora_B.default.weight": "model-00002-of-00004.safetensors",
80
+ "mlm.base_model.model.model.transformer.blocks.11.ff_norm.weight": "model-00002-of-00004.safetensors",
81
+ "mlm.base_model.model.model.transformer.blocks.11.ff_out.base_layer.weight": "model-00002-of-00004.safetensors",
82
+ "mlm.base_model.model.model.transformer.blocks.11.ff_out.lora_A.default.weight": "model-00002-of-00004.safetensors",
83
+ "mlm.base_model.model.model.transformer.blocks.11.ff_out.lora_B.default.weight": "model-00002-of-00004.safetensors",
84
+ "mlm.base_model.model.model.transformer.blocks.11.ff_proj.base_layer.weight": "model-00002-of-00004.safetensors",
85
+ "mlm.base_model.model.model.transformer.blocks.11.ff_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
86
+ "mlm.base_model.model.model.transformer.blocks.11.ff_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
87
+ "mlm.base_model.model.model.transformer.blocks.11.k_proj.base_layer.weight": "model-00002-of-00004.safetensors",
88
+ "mlm.base_model.model.model.transformer.blocks.11.k_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
89
+ "mlm.base_model.model.model.transformer.blocks.11.k_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
90
+ "mlm.base_model.model.model.transformer.blocks.11.q_proj.base_layer.weight": "model-00002-of-00004.safetensors",
91
+ "mlm.base_model.model.model.transformer.blocks.11.q_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
92
+ "mlm.base_model.model.model.transformer.blocks.11.q_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
93
+ "mlm.base_model.model.model.transformer.blocks.11.up_proj.base_layer.weight": "model-00002-of-00004.safetensors",
94
+ "mlm.base_model.model.model.transformer.blocks.11.up_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
95
+ "mlm.base_model.model.model.transformer.blocks.11.up_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
96
+ "mlm.base_model.model.model.transformer.blocks.11.v_proj.base_layer.weight": "model-00002-of-00004.safetensors",
97
+ "mlm.base_model.model.model.transformer.blocks.11.v_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
98
+ "mlm.base_model.model.model.transformer.blocks.11.v_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
99
+ "mlm.base_model.model.model.transformer.blocks.12.attn_norm.weight": "model-00002-of-00004.safetensors",
100
+ "mlm.base_model.model.model.transformer.blocks.12.attn_out.base_layer.weight": "model-00002-of-00004.safetensors",
101
+ "mlm.base_model.model.model.transformer.blocks.12.attn_out.lora_A.default.weight": "model-00002-of-00004.safetensors",
102
+ "mlm.base_model.model.model.transformer.blocks.12.attn_out.lora_B.default.weight": "model-00002-of-00004.safetensors",
103
+ "mlm.base_model.model.model.transformer.blocks.12.ff_norm.weight": "model-00002-of-00004.safetensors",
104
+ "mlm.base_model.model.model.transformer.blocks.12.ff_out.base_layer.weight": "model-00002-of-00004.safetensors",
105
+ "mlm.base_model.model.model.transformer.blocks.12.ff_out.lora_A.default.weight": "model-00002-of-00004.safetensors",
106
+ "mlm.base_model.model.model.transformer.blocks.12.ff_out.lora_B.default.weight": "model-00002-of-00004.safetensors",
107
+ "mlm.base_model.model.model.transformer.blocks.12.ff_proj.base_layer.weight": "model-00002-of-00004.safetensors",
108
+ "mlm.base_model.model.model.transformer.blocks.12.ff_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
109
+ "mlm.base_model.model.model.transformer.blocks.12.ff_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
110
+ "mlm.base_model.model.model.transformer.blocks.12.k_proj.base_layer.weight": "model-00002-of-00004.safetensors",
111
+ "mlm.base_model.model.model.transformer.blocks.12.k_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
112
+ "mlm.base_model.model.model.transformer.blocks.12.k_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
113
+ "mlm.base_model.model.model.transformer.blocks.12.q_proj.base_layer.weight": "model-00002-of-00004.safetensors",
114
+ "mlm.base_model.model.model.transformer.blocks.12.q_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
115
+ "mlm.base_model.model.model.transformer.blocks.12.q_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
116
+ "mlm.base_model.model.model.transformer.blocks.12.up_proj.base_layer.weight": "model-00002-of-00004.safetensors",
117
+ "mlm.base_model.model.model.transformer.blocks.12.up_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
118
+ "mlm.base_model.model.model.transformer.blocks.12.up_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
119
+ "mlm.base_model.model.model.transformer.blocks.12.v_proj.base_layer.weight": "model-00002-of-00004.safetensors",
120
+ "mlm.base_model.model.model.transformer.blocks.12.v_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
121
+ "mlm.base_model.model.model.transformer.blocks.12.v_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
122
+ "mlm.base_model.model.model.transformer.blocks.13.attn_norm.weight": "model-00002-of-00004.safetensors",
123
+ "mlm.base_model.model.model.transformer.blocks.13.attn_out.base_layer.weight": "model-00002-of-00004.safetensors",
124
+ "mlm.base_model.model.model.transformer.blocks.13.attn_out.lora_A.default.weight": "model-00002-of-00004.safetensors",
125
+ "mlm.base_model.model.model.transformer.blocks.13.attn_out.lora_B.default.weight": "model-00002-of-00004.safetensors",
126
+ "mlm.base_model.model.model.transformer.blocks.13.ff_norm.weight": "model-00002-of-00004.safetensors",
127
+ "mlm.base_model.model.model.transformer.blocks.13.ff_out.base_layer.weight": "model-00002-of-00004.safetensors",
128
+ "mlm.base_model.model.model.transformer.blocks.13.ff_out.lora_A.default.weight": "model-00002-of-00004.safetensors",
129
+ "mlm.base_model.model.model.transformer.blocks.13.ff_out.lora_B.default.weight": "model-00002-of-00004.safetensors",
130
+ "mlm.base_model.model.model.transformer.blocks.13.ff_proj.base_layer.weight": "model-00002-of-00004.safetensors",
131
+ "mlm.base_model.model.model.transformer.blocks.13.ff_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
132
+ "mlm.base_model.model.model.transformer.blocks.13.ff_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
133
+ "mlm.base_model.model.model.transformer.blocks.13.k_proj.base_layer.weight": "model-00002-of-00004.safetensors",
134
+ "mlm.base_model.model.model.transformer.blocks.13.k_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
135
+ "mlm.base_model.model.model.transformer.blocks.13.k_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
136
+ "mlm.base_model.model.model.transformer.blocks.13.q_proj.base_layer.weight": "model-00002-of-00004.safetensors",
137
+ "mlm.base_model.model.model.transformer.blocks.13.q_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
138
+ "mlm.base_model.model.model.transformer.blocks.13.q_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
139
+ "mlm.base_model.model.model.transformer.blocks.13.up_proj.base_layer.weight": "model-00002-of-00004.safetensors",
140
+ "mlm.base_model.model.model.transformer.blocks.13.up_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
141
+ "mlm.base_model.model.model.transformer.blocks.13.up_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
142
+ "mlm.base_model.model.model.transformer.blocks.13.v_proj.base_layer.weight": "model-00002-of-00004.safetensors",
143
+ "mlm.base_model.model.model.transformer.blocks.13.v_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
144
+ "mlm.base_model.model.model.transformer.blocks.13.v_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
145
+ "mlm.base_model.model.model.transformer.blocks.14.attn_norm.weight": "model-00002-of-00004.safetensors",
146
+ "mlm.base_model.model.model.transformer.blocks.14.attn_out.base_layer.weight": "model-00002-of-00004.safetensors",
147
+ "mlm.base_model.model.model.transformer.blocks.14.attn_out.lora_A.default.weight": "model-00002-of-00004.safetensors",
148
+ "mlm.base_model.model.model.transformer.blocks.14.attn_out.lora_B.default.weight": "model-00002-of-00004.safetensors",
149
+ "mlm.base_model.model.model.transformer.blocks.14.ff_norm.weight": "model-00002-of-00004.safetensors",
150
+ "mlm.base_model.model.model.transformer.blocks.14.ff_out.base_layer.weight": "model-00002-of-00004.safetensors",
151
+ "mlm.base_model.model.model.transformer.blocks.14.ff_out.lora_A.default.weight": "model-00002-of-00004.safetensors",
152
+ "mlm.base_model.model.model.transformer.blocks.14.ff_out.lora_B.default.weight": "model-00002-of-00004.safetensors",
153
+ "mlm.base_model.model.model.transformer.blocks.14.ff_proj.base_layer.weight": "model-00002-of-00004.safetensors",
154
+ "mlm.base_model.model.model.transformer.blocks.14.ff_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
155
+ "mlm.base_model.model.model.transformer.blocks.14.ff_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
156
+ "mlm.base_model.model.model.transformer.blocks.14.k_proj.base_layer.weight": "model-00002-of-00004.safetensors",
157
+ "mlm.base_model.model.model.transformer.blocks.14.k_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
158
+ "mlm.base_model.model.model.transformer.blocks.14.k_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
159
+ "mlm.base_model.model.model.transformer.blocks.14.q_proj.base_layer.weight": "model-00002-of-00004.safetensors",
160
+ "mlm.base_model.model.model.transformer.blocks.14.q_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
161
+ "mlm.base_model.model.model.transformer.blocks.14.q_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
162
+ "mlm.base_model.model.model.transformer.blocks.14.up_proj.base_layer.weight": "model-00002-of-00004.safetensors",
163
+ "mlm.base_model.model.model.transformer.blocks.14.up_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
164
+ "mlm.base_model.model.model.transformer.blocks.14.up_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
165
+ "mlm.base_model.model.model.transformer.blocks.14.v_proj.base_layer.weight": "model-00002-of-00004.safetensors",
166
+ "mlm.base_model.model.model.transformer.blocks.14.v_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
167
+ "mlm.base_model.model.model.transformer.blocks.14.v_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
168
+ "mlm.base_model.model.model.transformer.blocks.15.attn_norm.weight": "model-00002-of-00004.safetensors",
169
+ "mlm.base_model.model.model.transformer.blocks.15.attn_out.base_layer.weight": "model-00002-of-00004.safetensors",
170
+ "mlm.base_model.model.model.transformer.blocks.15.attn_out.lora_A.default.weight": "model-00002-of-00004.safetensors",
171
+ "mlm.base_model.model.model.transformer.blocks.15.attn_out.lora_B.default.weight": "model-00002-of-00004.safetensors",
172
+ "mlm.base_model.model.model.transformer.blocks.15.ff_norm.weight": "model-00002-of-00004.safetensors",
173
+ "mlm.base_model.model.model.transformer.blocks.15.ff_out.base_layer.weight": "model-00002-of-00004.safetensors",
174
+ "mlm.base_model.model.model.transformer.blocks.15.ff_out.lora_A.default.weight": "model-00002-of-00004.safetensors",
175
+ "mlm.base_model.model.model.transformer.blocks.15.ff_out.lora_B.default.weight": "model-00002-of-00004.safetensors",
176
+ "mlm.base_model.model.model.transformer.blocks.15.ff_proj.base_layer.weight": "model-00002-of-00004.safetensors",
177
+ "mlm.base_model.model.model.transformer.blocks.15.ff_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
178
+ "mlm.base_model.model.model.transformer.blocks.15.ff_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
179
+ "mlm.base_model.model.model.transformer.blocks.15.k_proj.base_layer.weight": "model-00002-of-00004.safetensors",
180
+ "mlm.base_model.model.model.transformer.blocks.15.k_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
181
+ "mlm.base_model.model.model.transformer.blocks.15.k_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
182
+ "mlm.base_model.model.model.transformer.blocks.15.q_proj.base_layer.weight": "model-00002-of-00004.safetensors",
183
+ "mlm.base_model.model.model.transformer.blocks.15.q_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
184
+ "mlm.base_model.model.model.transformer.blocks.15.q_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
185
+ "mlm.base_model.model.model.transformer.blocks.15.up_proj.base_layer.weight": "model-00002-of-00004.safetensors",
186
+ "mlm.base_model.model.model.transformer.blocks.15.up_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
187
+ "mlm.base_model.model.model.transformer.blocks.15.up_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
188
+ "mlm.base_model.model.model.transformer.blocks.15.v_proj.base_layer.weight": "model-00002-of-00004.safetensors",
189
+ "mlm.base_model.model.model.transformer.blocks.15.v_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
190
+ "mlm.base_model.model.model.transformer.blocks.15.v_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
191
+ "mlm.base_model.model.model.transformer.blocks.16.attn_norm.weight": "model-00002-of-00004.safetensors",
192
+ "mlm.base_model.model.model.transformer.blocks.16.attn_out.base_layer.weight": "model-00002-of-00004.safetensors",
193
+ "mlm.base_model.model.model.transformer.blocks.16.attn_out.lora_A.default.weight": "model-00002-of-00004.safetensors",
194
+ "mlm.base_model.model.model.transformer.blocks.16.attn_out.lora_B.default.weight": "model-00002-of-00004.safetensors",
195
+ "mlm.base_model.model.model.transformer.blocks.16.ff_norm.weight": "model-00002-of-00004.safetensors",
196
+ "mlm.base_model.model.model.transformer.blocks.16.ff_out.base_layer.weight": "model-00002-of-00004.safetensors",
197
+ "mlm.base_model.model.model.transformer.blocks.16.ff_out.lora_A.default.weight": "model-00002-of-00004.safetensors",
198
+ "mlm.base_model.model.model.transformer.blocks.16.ff_out.lora_B.default.weight": "model-00002-of-00004.safetensors",
199
+ "mlm.base_model.model.model.transformer.blocks.16.ff_proj.base_layer.weight": "model-00002-of-00004.safetensors",
200
+ "mlm.base_model.model.model.transformer.blocks.16.ff_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
201
+ "mlm.base_model.model.model.transformer.blocks.16.ff_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
202
+ "mlm.base_model.model.model.transformer.blocks.16.k_proj.base_layer.weight": "model-00002-of-00004.safetensors",
203
+ "mlm.base_model.model.model.transformer.blocks.16.k_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
204
+ "mlm.base_model.model.model.transformer.blocks.16.k_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
205
+ "mlm.base_model.model.model.transformer.blocks.16.q_proj.base_layer.weight": "model-00002-of-00004.safetensors",
206
+ "mlm.base_model.model.model.transformer.blocks.16.q_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
207
+ "mlm.base_model.model.model.transformer.blocks.16.q_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
208
+ "mlm.base_model.model.model.transformer.blocks.16.up_proj.base_layer.weight": "model-00002-of-00004.safetensors",
209
+ "mlm.base_model.model.model.transformer.blocks.16.up_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
210
+ "mlm.base_model.model.model.transformer.blocks.16.up_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
211
+ "mlm.base_model.model.model.transformer.blocks.16.v_proj.base_layer.weight": "model-00002-of-00004.safetensors",
212
+ "mlm.base_model.model.model.transformer.blocks.16.v_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
213
+ "mlm.base_model.model.model.transformer.blocks.16.v_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
214
+ "mlm.base_model.model.model.transformer.blocks.17.attn_norm.weight": "model-00002-of-00004.safetensors",
215
+ "mlm.base_model.model.model.transformer.blocks.17.attn_out.base_layer.weight": "model-00002-of-00004.safetensors",
216
+ "mlm.base_model.model.model.transformer.blocks.17.attn_out.lora_A.default.weight": "model-00002-of-00004.safetensors",
217
+ "mlm.base_model.model.model.transformer.blocks.17.attn_out.lora_B.default.weight": "model-00002-of-00004.safetensors",
218
+ "mlm.base_model.model.model.transformer.blocks.17.ff_norm.weight": "model-00002-of-00004.safetensors",
219
+ "mlm.base_model.model.model.transformer.blocks.17.ff_out.base_layer.weight": "model-00002-of-00004.safetensors",
220
+ "mlm.base_model.model.model.transformer.blocks.17.ff_out.lora_A.default.weight": "model-00002-of-00004.safetensors",
221
+ "mlm.base_model.model.model.transformer.blocks.17.ff_out.lora_B.default.weight": "model-00002-of-00004.safetensors",
222
+ "mlm.base_model.model.model.transformer.blocks.17.ff_proj.base_layer.weight": "model-00002-of-00004.safetensors",
223
+ "mlm.base_model.model.model.transformer.blocks.17.ff_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
224
+ "mlm.base_model.model.model.transformer.blocks.17.ff_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
225
+ "mlm.base_model.model.model.transformer.blocks.17.k_proj.base_layer.weight": "model-00002-of-00004.safetensors",
226
+ "mlm.base_model.model.model.transformer.blocks.17.k_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
227
+ "mlm.base_model.model.model.transformer.blocks.17.k_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
228
+ "mlm.base_model.model.model.transformer.blocks.17.q_proj.base_layer.weight": "model-00002-of-00004.safetensors",
229
+ "mlm.base_model.model.model.transformer.blocks.17.q_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
230
+ "mlm.base_model.model.model.transformer.blocks.17.q_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
231
+ "mlm.base_model.model.model.transformer.blocks.17.up_proj.base_layer.weight": "model-00002-of-00004.safetensors",
232
+ "mlm.base_model.model.model.transformer.blocks.17.up_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
233
+ "mlm.base_model.model.model.transformer.blocks.17.up_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
234
+ "mlm.base_model.model.model.transformer.blocks.17.v_proj.base_layer.weight": "model-00002-of-00004.safetensors",
235
+ "mlm.base_model.model.model.transformer.blocks.17.v_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
236
+ "mlm.base_model.model.model.transformer.blocks.17.v_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
237
+ "mlm.base_model.model.model.transformer.blocks.18.attn_norm.weight": "model-00002-of-00004.safetensors",
238
+ "mlm.base_model.model.model.transformer.blocks.18.attn_out.base_layer.weight": "model-00002-of-00004.safetensors",
239
+ "mlm.base_model.model.model.transformer.blocks.18.attn_out.lora_A.default.weight": "model-00002-of-00004.safetensors",
240
+ "mlm.base_model.model.model.transformer.blocks.18.attn_out.lora_B.default.weight": "model-00002-of-00004.safetensors",
241
+ "mlm.base_model.model.model.transformer.blocks.18.ff_norm.weight": "model-00002-of-00004.safetensors",
242
+ "mlm.base_model.model.model.transformer.blocks.18.ff_out.base_layer.weight": "model-00002-of-00004.safetensors",
243
+ "mlm.base_model.model.model.transformer.blocks.18.ff_out.lora_A.default.weight": "model-00002-of-00004.safetensors",
244
+ "mlm.base_model.model.model.transformer.blocks.18.ff_out.lora_B.default.weight": "model-00002-of-00004.safetensors",
245
+ "mlm.base_model.model.model.transformer.blocks.18.ff_proj.base_layer.weight": "model-00002-of-00004.safetensors",
246
+ "mlm.base_model.model.model.transformer.blocks.18.ff_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
247
+ "mlm.base_model.model.model.transformer.blocks.18.ff_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
248
+ "mlm.base_model.model.model.transformer.blocks.18.k_proj.base_layer.weight": "model-00002-of-00004.safetensors",
249
+ "mlm.base_model.model.model.transformer.blocks.18.k_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
250
+ "mlm.base_model.model.model.transformer.blocks.18.k_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
251
+ "mlm.base_model.model.model.transformer.blocks.18.q_proj.base_layer.weight": "model-00002-of-00004.safetensors",
252
+ "mlm.base_model.model.model.transformer.blocks.18.q_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
253
+ "mlm.base_model.model.model.transformer.blocks.18.q_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
254
+ "mlm.base_model.model.model.transformer.blocks.18.up_proj.base_layer.weight": "model-00002-of-00004.safetensors",
255
+ "mlm.base_model.model.model.transformer.blocks.18.up_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
256
+ "mlm.base_model.model.model.transformer.blocks.18.up_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
257
+ "mlm.base_model.model.model.transformer.blocks.18.v_proj.base_layer.weight": "model-00002-of-00004.safetensors",
258
+ "mlm.base_model.model.model.transformer.blocks.18.v_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
259
+ "mlm.base_model.model.model.transformer.blocks.18.v_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
260
+ "mlm.base_model.model.model.transformer.blocks.19.attn_norm.weight": "model-00003-of-00004.safetensors",
261
+ "mlm.base_model.model.model.transformer.blocks.19.attn_out.base_layer.weight": "model-00002-of-00004.safetensors",
262
+ "mlm.base_model.model.model.transformer.blocks.19.attn_out.lora_A.default.weight": "model-00002-of-00004.safetensors",
263
+ "mlm.base_model.model.model.transformer.blocks.19.attn_out.lora_B.default.weight": "model-00002-of-00004.safetensors",
264
+ "mlm.base_model.model.model.transformer.blocks.19.ff_norm.weight": "model-00003-of-00004.safetensors",
265
+ "mlm.base_model.model.model.transformer.blocks.19.ff_out.base_layer.weight": "model-00003-of-00004.safetensors",
266
+ "mlm.base_model.model.model.transformer.blocks.19.ff_out.lora_A.default.weight": "model-00003-of-00004.safetensors",
267
+ "mlm.base_model.model.model.transformer.blocks.19.ff_out.lora_B.default.weight": "model-00003-of-00004.safetensors",
268
+ "mlm.base_model.model.model.transformer.blocks.19.ff_proj.base_layer.weight": "model-00003-of-00004.safetensors",
269
+ "mlm.base_model.model.model.transformer.blocks.19.ff_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
270
+ "mlm.base_model.model.model.transformer.blocks.19.ff_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
271
+ "mlm.base_model.model.model.transformer.blocks.19.k_proj.base_layer.weight": "model-00003-of-00004.safetensors",
272
+ "mlm.base_model.model.model.transformer.blocks.19.k_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
273
+ "mlm.base_model.model.model.transformer.blocks.19.k_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
274
+ "mlm.base_model.model.model.transformer.blocks.19.q_proj.base_layer.weight": "model-00003-of-00004.safetensors",
275
+ "mlm.base_model.model.model.transformer.blocks.19.q_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
276
+ "mlm.base_model.model.model.transformer.blocks.19.q_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
277
+ "mlm.base_model.model.model.transformer.blocks.19.up_proj.base_layer.weight": "model-00003-of-00004.safetensors",
278
+ "mlm.base_model.model.model.transformer.blocks.19.up_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
279
+ "mlm.base_model.model.model.transformer.blocks.19.up_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
280
+ "mlm.base_model.model.model.transformer.blocks.19.v_proj.base_layer.weight": "model-00003-of-00004.safetensors",
281
+ "mlm.base_model.model.model.transformer.blocks.19.v_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
282
+ "mlm.base_model.model.model.transformer.blocks.19.v_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
283
+ "mlm.base_model.model.model.transformer.blocks.2.attn_norm.weight": "model-00001-of-00004.safetensors",
284
+ "mlm.base_model.model.model.transformer.blocks.2.attn_out.base_layer.weight": "model-00001-of-00004.safetensors",
285
+ "mlm.base_model.model.model.transformer.blocks.2.attn_out.lora_A.default.weight": "model-00001-of-00004.safetensors",
286
+ "mlm.base_model.model.model.transformer.blocks.2.attn_out.lora_B.default.weight": "model-00001-of-00004.safetensors",
287
+ "mlm.base_model.model.model.transformer.blocks.2.ff_norm.weight": "model-00001-of-00004.safetensors",
288
+ "mlm.base_model.model.model.transformer.blocks.2.ff_out.base_layer.weight": "model-00001-of-00004.safetensors",
289
+ "mlm.base_model.model.model.transformer.blocks.2.ff_out.lora_A.default.weight": "model-00001-of-00004.safetensors",
290
+ "mlm.base_model.model.model.transformer.blocks.2.ff_out.lora_B.default.weight": "model-00001-of-00004.safetensors",
291
+ "mlm.base_model.model.model.transformer.blocks.2.ff_proj.base_layer.weight": "model-00001-of-00004.safetensors",
292
+ "mlm.base_model.model.model.transformer.blocks.2.ff_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
293
+ "mlm.base_model.model.model.transformer.blocks.2.ff_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
294
+ "mlm.base_model.model.model.transformer.blocks.2.k_proj.base_layer.weight": "model-00001-of-00004.safetensors",
295
+ "mlm.base_model.model.model.transformer.blocks.2.k_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
296
+ "mlm.base_model.model.model.transformer.blocks.2.k_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
297
+ "mlm.base_model.model.model.transformer.blocks.2.q_proj.base_layer.weight": "model-00001-of-00004.safetensors",
298
+ "mlm.base_model.model.model.transformer.blocks.2.q_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
299
+ "mlm.base_model.model.model.transformer.blocks.2.q_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
300
+ "mlm.base_model.model.model.transformer.blocks.2.up_proj.base_layer.weight": "model-00001-of-00004.safetensors",
301
+ "mlm.base_model.model.model.transformer.blocks.2.up_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
302
+ "mlm.base_model.model.model.transformer.blocks.2.up_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
303
+ "mlm.base_model.model.model.transformer.blocks.2.v_proj.base_layer.weight": "model-00001-of-00004.safetensors",
304
+ "mlm.base_model.model.model.transformer.blocks.2.v_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
305
+ "mlm.base_model.model.model.transformer.blocks.2.v_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
306
+ "mlm.base_model.model.model.transformer.blocks.20.attn_norm.weight": "model-00003-of-00004.safetensors",
307
+ "mlm.base_model.model.model.transformer.blocks.20.attn_out.base_layer.weight": "model-00003-of-00004.safetensors",
308
+ "mlm.base_model.model.model.transformer.blocks.20.attn_out.lora_A.default.weight": "model-00003-of-00004.safetensors",
309
+ "mlm.base_model.model.model.transformer.blocks.20.attn_out.lora_B.default.weight": "model-00003-of-00004.safetensors",
310
+ "mlm.base_model.model.model.transformer.blocks.20.ff_norm.weight": "model-00003-of-00004.safetensors",
311
+ "mlm.base_model.model.model.transformer.blocks.20.ff_out.base_layer.weight": "model-00003-of-00004.safetensors",
312
+ "mlm.base_model.model.model.transformer.blocks.20.ff_out.lora_A.default.weight": "model-00003-of-00004.safetensors",
313
+ "mlm.base_model.model.model.transformer.blocks.20.ff_out.lora_B.default.weight": "model-00003-of-00004.safetensors",
314
+ "mlm.base_model.model.model.transformer.blocks.20.ff_proj.base_layer.weight": "model-00003-of-00004.safetensors",
315
+ "mlm.base_model.model.model.transformer.blocks.20.ff_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
316
+ "mlm.base_model.model.model.transformer.blocks.20.ff_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
317
+ "mlm.base_model.model.model.transformer.blocks.20.k_proj.base_layer.weight": "model-00003-of-00004.safetensors",
318
+ "mlm.base_model.model.model.transformer.blocks.20.k_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
319
+ "mlm.base_model.model.model.transformer.blocks.20.k_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
320
+ "mlm.base_model.model.model.transformer.blocks.20.q_proj.base_layer.weight": "model-00003-of-00004.safetensors",
321
+ "mlm.base_model.model.model.transformer.blocks.20.q_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
322
+ "mlm.base_model.model.model.transformer.blocks.20.q_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
323
+ "mlm.base_model.model.model.transformer.blocks.20.up_proj.base_layer.weight": "model-00003-of-00004.safetensors",
324
+ "mlm.base_model.model.model.transformer.blocks.20.up_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
325
+ "mlm.base_model.model.model.transformer.blocks.20.up_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
326
+ "mlm.base_model.model.model.transformer.blocks.20.v_proj.base_layer.weight": "model-00003-of-00004.safetensors",
327
+ "mlm.base_model.model.model.transformer.blocks.20.v_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
328
+ "mlm.base_model.model.model.transformer.blocks.20.v_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
329
+ "mlm.base_model.model.model.transformer.blocks.21.attn_norm.weight": "model-00003-of-00004.safetensors",
330
+ "mlm.base_model.model.model.transformer.blocks.21.attn_out.base_layer.weight": "model-00003-of-00004.safetensors",
331
+ "mlm.base_model.model.model.transformer.blocks.21.attn_out.lora_A.default.weight": "model-00003-of-00004.safetensors",
332
+ "mlm.base_model.model.model.transformer.blocks.21.attn_out.lora_B.default.weight": "model-00003-of-00004.safetensors",
333
+ "mlm.base_model.model.model.transformer.blocks.21.ff_norm.weight": "model-00003-of-00004.safetensors",
334
+ "mlm.base_model.model.model.transformer.blocks.21.ff_out.base_layer.weight": "model-00003-of-00004.safetensors",
335
+ "mlm.base_model.model.model.transformer.blocks.21.ff_out.lora_A.default.weight": "model-00003-of-00004.safetensors",
336
+ "mlm.base_model.model.model.transformer.blocks.21.ff_out.lora_B.default.weight": "model-00003-of-00004.safetensors",
337
+ "mlm.base_model.model.model.transformer.blocks.21.ff_proj.base_layer.weight": "model-00003-of-00004.safetensors",
338
+ "mlm.base_model.model.model.transformer.blocks.21.ff_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
339
+ "mlm.base_model.model.model.transformer.blocks.21.ff_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
340
+ "mlm.base_model.model.model.transformer.blocks.21.k_proj.base_layer.weight": "model-00003-of-00004.safetensors",
341
+ "mlm.base_model.model.model.transformer.blocks.21.k_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
342
+ "mlm.base_model.model.model.transformer.blocks.21.k_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
343
+ "mlm.base_model.model.model.transformer.blocks.21.q_proj.base_layer.weight": "model-00003-of-00004.safetensors",
344
+ "mlm.base_model.model.model.transformer.blocks.21.q_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
345
+ "mlm.base_model.model.model.transformer.blocks.21.q_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
346
+ "mlm.base_model.model.model.transformer.blocks.21.up_proj.base_layer.weight": "model-00003-of-00004.safetensors",
347
+ "mlm.base_model.model.model.transformer.blocks.21.up_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
348
+ "mlm.base_model.model.model.transformer.blocks.21.up_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
349
+ "mlm.base_model.model.model.transformer.blocks.21.v_proj.base_layer.weight": "model-00003-of-00004.safetensors",
350
+ "mlm.base_model.model.model.transformer.blocks.21.v_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
351
+ "mlm.base_model.model.model.transformer.blocks.21.v_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
352
+ "mlm.base_model.model.model.transformer.blocks.22.attn_norm.weight": "model-00003-of-00004.safetensors",
353
+ "mlm.base_model.model.model.transformer.blocks.22.attn_out.base_layer.weight": "model-00003-of-00004.safetensors",
354
+ "mlm.base_model.model.model.transformer.blocks.22.attn_out.lora_A.default.weight": "model-00003-of-00004.safetensors",
355
+ "mlm.base_model.model.model.transformer.blocks.22.attn_out.lora_B.default.weight": "model-00003-of-00004.safetensors",
356
+ "mlm.base_model.model.model.transformer.blocks.22.ff_norm.weight": "model-00003-of-00004.safetensors",
357
+ "mlm.base_model.model.model.transformer.blocks.22.ff_out.base_layer.weight": "model-00003-of-00004.safetensors",
358
+ "mlm.base_model.model.model.transformer.blocks.22.ff_out.lora_A.default.weight": "model-00003-of-00004.safetensors",
359
+ "mlm.base_model.model.model.transformer.blocks.22.ff_out.lora_B.default.weight": "model-00003-of-00004.safetensors",
360
+ "mlm.base_model.model.model.transformer.blocks.22.ff_proj.base_layer.weight": "model-00003-of-00004.safetensors",
361
+ "mlm.base_model.model.model.transformer.blocks.22.ff_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
362
+ "mlm.base_model.model.model.transformer.blocks.22.ff_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
363
+ "mlm.base_model.model.model.transformer.blocks.22.k_proj.base_layer.weight": "model-00003-of-00004.safetensors",
364
+ "mlm.base_model.model.model.transformer.blocks.22.k_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
365
+ "mlm.base_model.model.model.transformer.blocks.22.k_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
366
+ "mlm.base_model.model.model.transformer.blocks.22.q_proj.base_layer.weight": "model-00003-of-00004.safetensors",
367
+ "mlm.base_model.model.model.transformer.blocks.22.q_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
368
+ "mlm.base_model.model.model.transformer.blocks.22.q_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
369
+ "mlm.base_model.model.model.transformer.blocks.22.up_proj.base_layer.weight": "model-00003-of-00004.safetensors",
370
+ "mlm.base_model.model.model.transformer.blocks.22.up_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
371
+ "mlm.base_model.model.model.transformer.blocks.22.up_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
372
+ "mlm.base_model.model.model.transformer.blocks.22.v_proj.base_layer.weight": "model-00003-of-00004.safetensors",
373
+ "mlm.base_model.model.model.transformer.blocks.22.v_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
374
+ "mlm.base_model.model.model.transformer.blocks.22.v_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
375
+ "mlm.base_model.model.model.transformer.blocks.23.attn_norm.weight": "model-00003-of-00004.safetensors",
376
+ "mlm.base_model.model.model.transformer.blocks.23.attn_out.base_layer.weight": "model-00003-of-00004.safetensors",
377
+ "mlm.base_model.model.model.transformer.blocks.23.attn_out.lora_A.default.weight": "model-00003-of-00004.safetensors",
378
+ "mlm.base_model.model.model.transformer.blocks.23.attn_out.lora_B.default.weight": "model-00003-of-00004.safetensors",
379
+ "mlm.base_model.model.model.transformer.blocks.23.ff_norm.weight": "model-00003-of-00004.safetensors",
380
+ "mlm.base_model.model.model.transformer.blocks.23.ff_out.base_layer.weight": "model-00003-of-00004.safetensors",
381
+ "mlm.base_model.model.model.transformer.blocks.23.ff_out.lora_A.default.weight": "model-00003-of-00004.safetensors",
382
+ "mlm.base_model.model.model.transformer.blocks.23.ff_out.lora_B.default.weight": "model-00003-of-00004.safetensors",
383
+ "mlm.base_model.model.model.transformer.blocks.23.ff_proj.base_layer.weight": "model-00003-of-00004.safetensors",
384
+ "mlm.base_model.model.model.transformer.blocks.23.ff_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
385
+ "mlm.base_model.model.model.transformer.blocks.23.ff_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
386
+ "mlm.base_model.model.model.transformer.blocks.23.k_proj.base_layer.weight": "model-00003-of-00004.safetensors",
387
+ "mlm.base_model.model.model.transformer.blocks.23.k_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
388
+ "mlm.base_model.model.model.transformer.blocks.23.k_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
389
+ "mlm.base_model.model.model.transformer.blocks.23.q_proj.base_layer.weight": "model-00003-of-00004.safetensors",
390
+ "mlm.base_model.model.model.transformer.blocks.23.q_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
391
+ "mlm.base_model.model.model.transformer.blocks.23.q_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
392
+ "mlm.base_model.model.model.transformer.blocks.23.up_proj.base_layer.weight": "model-00003-of-00004.safetensors",
393
+ "mlm.base_model.model.model.transformer.blocks.23.up_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
394
+ "mlm.base_model.model.model.transformer.blocks.23.up_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
395
+ "mlm.base_model.model.model.transformer.blocks.23.v_proj.base_layer.weight": "model-00003-of-00004.safetensors",
396
+ "mlm.base_model.model.model.transformer.blocks.23.v_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
397
+ "mlm.base_model.model.model.transformer.blocks.23.v_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
398
+ "mlm.base_model.model.model.transformer.blocks.24.attn_norm.weight": "model-00003-of-00004.safetensors",
399
+ "mlm.base_model.model.model.transformer.blocks.24.attn_out.base_layer.weight": "model-00003-of-00004.safetensors",
400
+ "mlm.base_model.model.model.transformer.blocks.24.attn_out.lora_A.default.weight": "model-00003-of-00004.safetensors",
401
+ "mlm.base_model.model.model.transformer.blocks.24.attn_out.lora_B.default.weight": "model-00003-of-00004.safetensors",
402
+ "mlm.base_model.model.model.transformer.blocks.24.ff_norm.weight": "model-00003-of-00004.safetensors",
403
+ "mlm.base_model.model.model.transformer.blocks.24.ff_out.base_layer.weight": "model-00003-of-00004.safetensors",
404
+ "mlm.base_model.model.model.transformer.blocks.24.ff_out.lora_A.default.weight": "model-00003-of-00004.safetensors",
405
+ "mlm.base_model.model.model.transformer.blocks.24.ff_out.lora_B.default.weight": "model-00003-of-00004.safetensors",
406
+ "mlm.base_model.model.model.transformer.blocks.24.ff_proj.base_layer.weight": "model-00003-of-00004.safetensors",
407
+ "mlm.base_model.model.model.transformer.blocks.24.ff_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
408
+ "mlm.base_model.model.model.transformer.blocks.24.ff_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
409
+ "mlm.base_model.model.model.transformer.blocks.24.k_proj.base_layer.weight": "model-00003-of-00004.safetensors",
410
+ "mlm.base_model.model.model.transformer.blocks.24.k_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
411
+ "mlm.base_model.model.model.transformer.blocks.24.k_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
412
+ "mlm.base_model.model.model.transformer.blocks.24.q_proj.base_layer.weight": "model-00003-of-00004.safetensors",
413
+ "mlm.base_model.model.model.transformer.blocks.24.q_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
414
+ "mlm.base_model.model.model.transformer.blocks.24.q_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
415
+ "mlm.base_model.model.model.transformer.blocks.24.up_proj.base_layer.weight": "model-00003-of-00004.safetensors",
416
+ "mlm.base_model.model.model.transformer.blocks.24.up_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
417
+ "mlm.base_model.model.model.transformer.blocks.24.up_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
418
+ "mlm.base_model.model.model.transformer.blocks.24.v_proj.base_layer.weight": "model-00003-of-00004.safetensors",
419
+ "mlm.base_model.model.model.transformer.blocks.24.v_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
420
+ "mlm.base_model.model.model.transformer.blocks.24.v_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
421
+ "mlm.base_model.model.model.transformer.blocks.25.attn_norm.weight": "model-00003-of-00004.safetensors",
422
+ "mlm.base_model.model.model.transformer.blocks.25.attn_out.base_layer.weight": "model-00003-of-00004.safetensors",
423
+ "mlm.base_model.model.model.transformer.blocks.25.attn_out.lora_A.default.weight": "model-00003-of-00004.safetensors",
424
+ "mlm.base_model.model.model.transformer.blocks.25.attn_out.lora_B.default.weight": "model-00003-of-00004.safetensors",
425
+ "mlm.base_model.model.model.transformer.blocks.25.ff_norm.weight": "model-00003-of-00004.safetensors",
426
+ "mlm.base_model.model.model.transformer.blocks.25.ff_out.base_layer.weight": "model-00003-of-00004.safetensors",
427
+ "mlm.base_model.model.model.transformer.blocks.25.ff_out.lora_A.default.weight": "model-00003-of-00004.safetensors",
428
+ "mlm.base_model.model.model.transformer.blocks.25.ff_out.lora_B.default.weight": "model-00003-of-00004.safetensors",
429
+ "mlm.base_model.model.model.transformer.blocks.25.ff_proj.base_layer.weight": "model-00003-of-00004.safetensors",
430
+ "mlm.base_model.model.model.transformer.blocks.25.ff_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
431
+ "mlm.base_model.model.model.transformer.blocks.25.ff_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
432
+ "mlm.base_model.model.model.transformer.blocks.25.k_proj.base_layer.weight": "model-00003-of-00004.safetensors",
433
+ "mlm.base_model.model.model.transformer.blocks.25.k_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
434
+ "mlm.base_model.model.model.transformer.blocks.25.k_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
435
+ "mlm.base_model.model.model.transformer.blocks.25.q_proj.base_layer.weight": "model-00003-of-00004.safetensors",
436
+ "mlm.base_model.model.model.transformer.blocks.25.q_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
437
+ "mlm.base_model.model.model.transformer.blocks.25.q_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
438
+ "mlm.base_model.model.model.transformer.blocks.25.up_proj.base_layer.weight": "model-00003-of-00004.safetensors",
439
+ "mlm.base_model.model.model.transformer.blocks.25.up_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
440
+ "mlm.base_model.model.model.transformer.blocks.25.up_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
441
+ "mlm.base_model.model.model.transformer.blocks.25.v_proj.base_layer.weight": "model-00003-of-00004.safetensors",
442
+ "mlm.base_model.model.model.transformer.blocks.25.v_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
443
+ "mlm.base_model.model.model.transformer.blocks.25.v_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
444
+ "mlm.base_model.model.model.transformer.blocks.26.attn_norm.weight": "model-00003-of-00004.safetensors",
445
+ "mlm.base_model.model.model.transformer.blocks.26.attn_out.base_layer.weight": "model-00003-of-00004.safetensors",
446
+ "mlm.base_model.model.model.transformer.blocks.26.attn_out.lora_A.default.weight": "model-00003-of-00004.safetensors",
447
+ "mlm.base_model.model.model.transformer.blocks.26.attn_out.lora_B.default.weight": "model-00003-of-00004.safetensors",
448
+ "mlm.base_model.model.model.transformer.blocks.26.ff_norm.weight": "model-00003-of-00004.safetensors",
449
+ "mlm.base_model.model.model.transformer.blocks.26.ff_out.base_layer.weight": "model-00003-of-00004.safetensors",
450
+ "mlm.base_model.model.model.transformer.blocks.26.ff_out.lora_A.default.weight": "model-00003-of-00004.safetensors",
451
+ "mlm.base_model.model.model.transformer.blocks.26.ff_out.lora_B.default.weight": "model-00003-of-00004.safetensors",
452
+ "mlm.base_model.model.model.transformer.blocks.26.ff_proj.base_layer.weight": "model-00003-of-00004.safetensors",
453
+ "mlm.base_model.model.model.transformer.blocks.26.ff_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
454
+ "mlm.base_model.model.model.transformer.blocks.26.ff_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
455
+ "mlm.base_model.model.model.transformer.blocks.26.k_proj.base_layer.weight": "model-00003-of-00004.safetensors",
456
+ "mlm.base_model.model.model.transformer.blocks.26.k_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
457
+ "mlm.base_model.model.model.transformer.blocks.26.k_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
458
+ "mlm.base_model.model.model.transformer.blocks.26.q_proj.base_layer.weight": "model-00003-of-00004.safetensors",
459
+ "mlm.base_model.model.model.transformer.blocks.26.q_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
460
+ "mlm.base_model.model.model.transformer.blocks.26.q_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
461
+ "mlm.base_model.model.model.transformer.blocks.26.up_proj.base_layer.weight": "model-00003-of-00004.safetensors",
462
+ "mlm.base_model.model.model.transformer.blocks.26.up_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
463
+ "mlm.base_model.model.model.transformer.blocks.26.up_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
464
+ "mlm.base_model.model.model.transformer.blocks.26.v_proj.base_layer.weight": "model-00003-of-00004.safetensors",
465
+ "mlm.base_model.model.model.transformer.blocks.26.v_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
466
+ "mlm.base_model.model.model.transformer.blocks.26.v_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
467
+ "mlm.base_model.model.model.transformer.blocks.27.attn_norm.weight": "model-00003-of-00004.safetensors",
468
+ "mlm.base_model.model.model.transformer.blocks.27.attn_out.base_layer.weight": "model-00003-of-00004.safetensors",
469
+ "mlm.base_model.model.model.transformer.blocks.27.attn_out.lora_A.default.weight": "model-00003-of-00004.safetensors",
470
+ "mlm.base_model.model.model.transformer.blocks.27.attn_out.lora_B.default.weight": "model-00003-of-00004.safetensors",
471
+ "mlm.base_model.model.model.transformer.blocks.27.ff_norm.weight": "model-00003-of-00004.safetensors",
472
+ "mlm.base_model.model.model.transformer.blocks.27.ff_out.base_layer.weight": "model-00003-of-00004.safetensors",
473
+ "mlm.base_model.model.model.transformer.blocks.27.ff_out.lora_A.default.weight": "model-00003-of-00004.safetensors",
474
+ "mlm.base_model.model.model.transformer.blocks.27.ff_out.lora_B.default.weight": "model-00003-of-00004.safetensors",
475
+ "mlm.base_model.model.model.transformer.blocks.27.ff_proj.base_layer.weight": "model-00003-of-00004.safetensors",
476
+ "mlm.base_model.model.model.transformer.blocks.27.ff_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
477
+ "mlm.base_model.model.model.transformer.blocks.27.ff_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
478
+ "mlm.base_model.model.model.transformer.blocks.27.k_proj.base_layer.weight": "model-00003-of-00004.safetensors",
479
+ "mlm.base_model.model.model.transformer.blocks.27.k_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
480
+ "mlm.base_model.model.model.transformer.blocks.27.k_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
481
+ "mlm.base_model.model.model.transformer.blocks.27.q_proj.base_layer.weight": "model-00003-of-00004.safetensors",
482
+ "mlm.base_model.model.model.transformer.blocks.27.q_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
483
+ "mlm.base_model.model.model.transformer.blocks.27.q_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
484
+ "mlm.base_model.model.model.transformer.blocks.27.up_proj.base_layer.weight": "model-00003-of-00004.safetensors",
485
+ "mlm.base_model.model.model.transformer.blocks.27.up_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
486
+ "mlm.base_model.model.model.transformer.blocks.27.up_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
487
+ "mlm.base_model.model.model.transformer.blocks.27.v_proj.base_layer.weight": "model-00003-of-00004.safetensors",
488
+ "mlm.base_model.model.model.transformer.blocks.27.v_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
489
+ "mlm.base_model.model.model.transformer.blocks.27.v_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
490
+ "mlm.base_model.model.model.transformer.blocks.28.attn_norm.weight": "model-00003-of-00004.safetensors",
491
+ "mlm.base_model.model.model.transformer.blocks.28.attn_out.base_layer.weight": "model-00003-of-00004.safetensors",
492
+ "mlm.base_model.model.model.transformer.blocks.28.attn_out.lora_A.default.weight": "model-00003-of-00004.safetensors",
493
+ "mlm.base_model.model.model.transformer.blocks.28.attn_out.lora_B.default.weight": "model-00003-of-00004.safetensors",
494
+ "mlm.base_model.model.model.transformer.blocks.28.ff_norm.weight": "model-00003-of-00004.safetensors",
495
+ "mlm.base_model.model.model.transformer.blocks.28.ff_out.base_layer.weight": "model-00003-of-00004.safetensors",
496
+ "mlm.base_model.model.model.transformer.blocks.28.ff_out.lora_A.default.weight": "model-00003-of-00004.safetensors",
497
+ "mlm.base_model.model.model.transformer.blocks.28.ff_out.lora_B.default.weight": "model-00003-of-00004.safetensors",
498
+ "mlm.base_model.model.model.transformer.blocks.28.ff_proj.base_layer.weight": "model-00004-of-00004.safetensors",
499
+ "mlm.base_model.model.model.transformer.blocks.28.ff_proj.lora_A.default.weight": "model-00004-of-00004.safetensors",
500
+ "mlm.base_model.model.model.transformer.blocks.28.ff_proj.lora_B.default.weight": "model-00004-of-00004.safetensors",
501
+ "mlm.base_model.model.model.transformer.blocks.28.k_proj.base_layer.weight": "model-00003-of-00004.safetensors",
502
+ "mlm.base_model.model.model.transformer.blocks.28.k_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
503
+ "mlm.base_model.model.model.transformer.blocks.28.k_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
504
+ "mlm.base_model.model.model.transformer.blocks.28.q_proj.base_layer.weight": "model-00003-of-00004.safetensors",
505
+ "mlm.base_model.model.model.transformer.blocks.28.q_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
506
+ "mlm.base_model.model.model.transformer.blocks.28.q_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
507
+ "mlm.base_model.model.model.transformer.blocks.28.up_proj.base_layer.weight": "model-00004-of-00004.safetensors",
508
+ "mlm.base_model.model.model.transformer.blocks.28.up_proj.lora_A.default.weight": "model-00004-of-00004.safetensors",
509
+ "mlm.base_model.model.model.transformer.blocks.28.up_proj.lora_B.default.weight": "model-00004-of-00004.safetensors",
510
+ "mlm.base_model.model.model.transformer.blocks.28.v_proj.base_layer.weight": "model-00003-of-00004.safetensors",
511
+ "mlm.base_model.model.model.transformer.blocks.28.v_proj.lora_A.default.weight": "model-00003-of-00004.safetensors",
512
+ "mlm.base_model.model.model.transformer.blocks.28.v_proj.lora_B.default.weight": "model-00003-of-00004.safetensors",
513
+ "mlm.base_model.model.model.transformer.blocks.29.attn_norm.weight": "model-00004-of-00004.safetensors",
514
+ "mlm.base_model.model.model.transformer.blocks.29.attn_out.base_layer.weight": "model-00004-of-00004.safetensors",
515
+ "mlm.base_model.model.model.transformer.blocks.29.attn_out.lora_A.default.weight": "model-00004-of-00004.safetensors",
516
+ "mlm.base_model.model.model.transformer.blocks.29.attn_out.lora_B.default.weight": "model-00004-of-00004.safetensors",
517
+ "mlm.base_model.model.model.transformer.blocks.29.ff_norm.weight": "model-00004-of-00004.safetensors",
518
+ "mlm.base_model.model.model.transformer.blocks.29.ff_out.base_layer.weight": "model-00004-of-00004.safetensors",
519
+ "mlm.base_model.model.model.transformer.blocks.29.ff_out.lora_A.default.weight": "model-00004-of-00004.safetensors",
520
+ "mlm.base_model.model.model.transformer.blocks.29.ff_out.lora_B.default.weight": "model-00004-of-00004.safetensors",
521
+ "mlm.base_model.model.model.transformer.blocks.29.ff_proj.base_layer.weight": "model-00004-of-00004.safetensors",
522
+ "mlm.base_model.model.model.transformer.blocks.29.ff_proj.lora_A.default.weight": "model-00004-of-00004.safetensors",
523
+ "mlm.base_model.model.model.transformer.blocks.29.ff_proj.lora_B.default.weight": "model-00004-of-00004.safetensors",
524
+ "mlm.base_model.model.model.transformer.blocks.29.k_proj.base_layer.weight": "model-00004-of-00004.safetensors",
525
+ "mlm.base_model.model.model.transformer.blocks.29.k_proj.lora_A.default.weight": "model-00004-of-00004.safetensors",
526
+ "mlm.base_model.model.model.transformer.blocks.29.k_proj.lora_B.default.weight": "model-00004-of-00004.safetensors",
527
+ "mlm.base_model.model.model.transformer.blocks.29.q_proj.base_layer.weight": "model-00004-of-00004.safetensors",
528
+ "mlm.base_model.model.model.transformer.blocks.29.q_proj.lora_A.default.weight": "model-00004-of-00004.safetensors",
529
+ "mlm.base_model.model.model.transformer.blocks.29.q_proj.lora_B.default.weight": "model-00004-of-00004.safetensors",
530
+ "mlm.base_model.model.model.transformer.blocks.29.up_proj.base_layer.weight": "model-00004-of-00004.safetensors",
531
+ "mlm.base_model.model.model.transformer.blocks.29.up_proj.lora_A.default.weight": "model-00004-of-00004.safetensors",
532
+ "mlm.base_model.model.model.transformer.blocks.29.up_proj.lora_B.default.weight": "model-00004-of-00004.safetensors",
533
+ "mlm.base_model.model.model.transformer.blocks.29.v_proj.base_layer.weight": "model-00004-of-00004.safetensors",
534
+ "mlm.base_model.model.model.transformer.blocks.29.v_proj.lora_A.default.weight": "model-00004-of-00004.safetensors",
535
+ "mlm.base_model.model.model.transformer.blocks.29.v_proj.lora_B.default.weight": "model-00004-of-00004.safetensors",
536
+ "mlm.base_model.model.model.transformer.blocks.3.attn_norm.weight": "model-00001-of-00004.safetensors",
537
+ "mlm.base_model.model.model.transformer.blocks.3.attn_out.base_layer.weight": "model-00001-of-00004.safetensors",
538
+ "mlm.base_model.model.model.transformer.blocks.3.attn_out.lora_A.default.weight": "model-00001-of-00004.safetensors",
539
+ "mlm.base_model.model.model.transformer.blocks.3.attn_out.lora_B.default.weight": "model-00001-of-00004.safetensors",
540
+ "mlm.base_model.model.model.transformer.blocks.3.ff_norm.weight": "model-00001-of-00004.safetensors",
541
+ "mlm.base_model.model.model.transformer.blocks.3.ff_out.base_layer.weight": "model-00001-of-00004.safetensors",
542
+ "mlm.base_model.model.model.transformer.blocks.3.ff_out.lora_A.default.weight": "model-00001-of-00004.safetensors",
543
+ "mlm.base_model.model.model.transformer.blocks.3.ff_out.lora_B.default.weight": "model-00001-of-00004.safetensors",
544
+ "mlm.base_model.model.model.transformer.blocks.3.ff_proj.base_layer.weight": "model-00001-of-00004.safetensors",
545
+ "mlm.base_model.model.model.transformer.blocks.3.ff_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
546
+ "mlm.base_model.model.model.transformer.blocks.3.ff_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
547
+ "mlm.base_model.model.model.transformer.blocks.3.k_proj.base_layer.weight": "model-00001-of-00004.safetensors",
548
+ "mlm.base_model.model.model.transformer.blocks.3.k_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
549
+ "mlm.base_model.model.model.transformer.blocks.3.k_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
550
+ "mlm.base_model.model.model.transformer.blocks.3.q_proj.base_layer.weight": "model-00001-of-00004.safetensors",
551
+ "mlm.base_model.model.model.transformer.blocks.3.q_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
552
+ "mlm.base_model.model.model.transformer.blocks.3.q_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
553
+ "mlm.base_model.model.model.transformer.blocks.3.up_proj.base_layer.weight": "model-00001-of-00004.safetensors",
554
+ "mlm.base_model.model.model.transformer.blocks.3.up_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
555
+ "mlm.base_model.model.model.transformer.blocks.3.up_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
556
+ "mlm.base_model.model.model.transformer.blocks.3.v_proj.base_layer.weight": "model-00001-of-00004.safetensors",
557
+ "mlm.base_model.model.model.transformer.blocks.3.v_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
558
+ "mlm.base_model.model.model.transformer.blocks.3.v_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
559
+ "mlm.base_model.model.model.transformer.blocks.30.attn_norm.weight": "model-00004-of-00004.safetensors",
560
+ "mlm.base_model.model.model.transformer.blocks.30.attn_out.base_layer.weight": "model-00004-of-00004.safetensors",
561
+ "mlm.base_model.model.model.transformer.blocks.30.attn_out.lora_A.default.weight": "model-00004-of-00004.safetensors",
562
+ "mlm.base_model.model.model.transformer.blocks.30.attn_out.lora_B.default.weight": "model-00004-of-00004.safetensors",
563
+ "mlm.base_model.model.model.transformer.blocks.30.ff_norm.weight": "model-00004-of-00004.safetensors",
564
+ "mlm.base_model.model.model.transformer.blocks.30.ff_out.base_layer.weight": "model-00004-of-00004.safetensors",
565
+ "mlm.base_model.model.model.transformer.blocks.30.ff_out.lora_A.default.weight": "model-00004-of-00004.safetensors",
566
+ "mlm.base_model.model.model.transformer.blocks.30.ff_out.lora_B.default.weight": "model-00004-of-00004.safetensors",
567
+ "mlm.base_model.model.model.transformer.blocks.30.ff_proj.base_layer.weight": "model-00004-of-00004.safetensors",
568
+ "mlm.base_model.model.model.transformer.blocks.30.ff_proj.lora_A.default.weight": "model-00004-of-00004.safetensors",
569
+ "mlm.base_model.model.model.transformer.blocks.30.ff_proj.lora_B.default.weight": "model-00004-of-00004.safetensors",
570
+ "mlm.base_model.model.model.transformer.blocks.30.k_proj.base_layer.weight": "model-00004-of-00004.safetensors",
571
+ "mlm.base_model.model.model.transformer.blocks.30.k_proj.lora_A.default.weight": "model-00004-of-00004.safetensors",
572
+ "mlm.base_model.model.model.transformer.blocks.30.k_proj.lora_B.default.weight": "model-00004-of-00004.safetensors",
573
+ "mlm.base_model.model.model.transformer.blocks.30.q_proj.base_layer.weight": "model-00004-of-00004.safetensors",
574
+ "mlm.base_model.model.model.transformer.blocks.30.q_proj.lora_A.default.weight": "model-00004-of-00004.safetensors",
575
+ "mlm.base_model.model.model.transformer.blocks.30.q_proj.lora_B.default.weight": "model-00004-of-00004.safetensors",
576
+ "mlm.base_model.model.model.transformer.blocks.30.up_proj.base_layer.weight": "model-00004-of-00004.safetensors",
577
+ "mlm.base_model.model.model.transformer.blocks.30.up_proj.lora_A.default.weight": "model-00004-of-00004.safetensors",
578
+ "mlm.base_model.model.model.transformer.blocks.30.up_proj.lora_B.default.weight": "model-00004-of-00004.safetensors",
579
+ "mlm.base_model.model.model.transformer.blocks.30.v_proj.base_layer.weight": "model-00004-of-00004.safetensors",
580
+ "mlm.base_model.model.model.transformer.blocks.30.v_proj.lora_A.default.weight": "model-00004-of-00004.safetensors",
581
+ "mlm.base_model.model.model.transformer.blocks.30.v_proj.lora_B.default.weight": "model-00004-of-00004.safetensors",
582
+ "mlm.base_model.model.model.transformer.blocks.31.attn_norm.weight": "model-00004-of-00004.safetensors",
583
+ "mlm.base_model.model.model.transformer.blocks.31.attn_out.base_layer.weight": "model-00004-of-00004.safetensors",
584
+ "mlm.base_model.model.model.transformer.blocks.31.attn_out.lora_A.default.weight": "model-00004-of-00004.safetensors",
585
+ "mlm.base_model.model.model.transformer.blocks.31.attn_out.lora_B.default.weight": "model-00004-of-00004.safetensors",
586
+ "mlm.base_model.model.model.transformer.blocks.31.ff_norm.weight": "model-00004-of-00004.safetensors",
587
+ "mlm.base_model.model.model.transformer.blocks.31.ff_out.base_layer.weight": "model-00004-of-00004.safetensors",
588
+ "mlm.base_model.model.model.transformer.blocks.31.ff_out.lora_A.default.weight": "model-00004-of-00004.safetensors",
589
+ "mlm.base_model.model.model.transformer.blocks.31.ff_out.lora_B.default.weight": "model-00004-of-00004.safetensors",
590
+ "mlm.base_model.model.model.transformer.blocks.31.ff_proj.base_layer.weight": "model-00004-of-00004.safetensors",
591
+ "mlm.base_model.model.model.transformer.blocks.31.ff_proj.lora_A.default.weight": "model-00004-of-00004.safetensors",
592
+ "mlm.base_model.model.model.transformer.blocks.31.ff_proj.lora_B.default.weight": "model-00004-of-00004.safetensors",
593
+ "mlm.base_model.model.model.transformer.blocks.31.k_proj.base_layer.weight": "model-00004-of-00004.safetensors",
594
+ "mlm.base_model.model.model.transformer.blocks.31.k_proj.lora_A.default.weight": "model-00004-of-00004.safetensors",
595
+ "mlm.base_model.model.model.transformer.blocks.31.k_proj.lora_B.default.weight": "model-00004-of-00004.safetensors",
596
+ "mlm.base_model.model.model.transformer.blocks.31.q_proj.base_layer.weight": "model-00004-of-00004.safetensors",
597
+ "mlm.base_model.model.model.transformer.blocks.31.q_proj.lora_A.default.weight": "model-00004-of-00004.safetensors",
598
+ "mlm.base_model.model.model.transformer.blocks.31.q_proj.lora_B.default.weight": "model-00004-of-00004.safetensors",
599
+ "mlm.base_model.model.model.transformer.blocks.31.up_proj.base_layer.weight": "model-00004-of-00004.safetensors",
600
+ "mlm.base_model.model.model.transformer.blocks.31.up_proj.lora_A.default.weight": "model-00004-of-00004.safetensors",
601
+ "mlm.base_model.model.model.transformer.blocks.31.up_proj.lora_B.default.weight": "model-00004-of-00004.safetensors",
602
+ "mlm.base_model.model.model.transformer.blocks.31.v_proj.base_layer.weight": "model-00004-of-00004.safetensors",
603
+ "mlm.base_model.model.model.transformer.blocks.31.v_proj.lora_A.default.weight": "model-00004-of-00004.safetensors",
604
+ "mlm.base_model.model.model.transformer.blocks.31.v_proj.lora_B.default.weight": "model-00004-of-00004.safetensors",
605
+ "mlm.base_model.model.model.transformer.blocks.4.attn_norm.weight": "model-00001-of-00004.safetensors",
606
+ "mlm.base_model.model.model.transformer.blocks.4.attn_out.base_layer.weight": "model-00001-of-00004.safetensors",
607
+ "mlm.base_model.model.model.transformer.blocks.4.attn_out.lora_A.default.weight": "model-00001-of-00004.safetensors",
608
+ "mlm.base_model.model.model.transformer.blocks.4.attn_out.lora_B.default.weight": "model-00001-of-00004.safetensors",
609
+ "mlm.base_model.model.model.transformer.blocks.4.ff_norm.weight": "model-00001-of-00004.safetensors",
610
+ "mlm.base_model.model.model.transformer.blocks.4.ff_out.base_layer.weight": "model-00001-of-00004.safetensors",
611
+ "mlm.base_model.model.model.transformer.blocks.4.ff_out.lora_A.default.weight": "model-00001-of-00004.safetensors",
612
+ "mlm.base_model.model.model.transformer.blocks.4.ff_out.lora_B.default.weight": "model-00001-of-00004.safetensors",
613
+ "mlm.base_model.model.model.transformer.blocks.4.ff_proj.base_layer.weight": "model-00001-of-00004.safetensors",
614
+ "mlm.base_model.model.model.transformer.blocks.4.ff_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
615
+ "mlm.base_model.model.model.transformer.blocks.4.ff_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
616
+ "mlm.base_model.model.model.transformer.blocks.4.k_proj.base_layer.weight": "model-00001-of-00004.safetensors",
617
+ "mlm.base_model.model.model.transformer.blocks.4.k_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
618
+ "mlm.base_model.model.model.transformer.blocks.4.k_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
619
+ "mlm.base_model.model.model.transformer.blocks.4.q_proj.base_layer.weight": "model-00001-of-00004.safetensors",
620
+ "mlm.base_model.model.model.transformer.blocks.4.q_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
621
+ "mlm.base_model.model.model.transformer.blocks.4.q_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
622
+ "mlm.base_model.model.model.transformer.blocks.4.up_proj.base_layer.weight": "model-00001-of-00004.safetensors",
623
+ "mlm.base_model.model.model.transformer.blocks.4.up_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
624
+ "mlm.base_model.model.model.transformer.blocks.4.up_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
625
+ "mlm.base_model.model.model.transformer.blocks.4.v_proj.base_layer.weight": "model-00001-of-00004.safetensors",
626
+ "mlm.base_model.model.model.transformer.blocks.4.v_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
627
+ "mlm.base_model.model.model.transformer.blocks.4.v_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
628
+ "mlm.base_model.model.model.transformer.blocks.5.attn_norm.weight": "model-00001-of-00004.safetensors",
629
+ "mlm.base_model.model.model.transformer.blocks.5.attn_out.base_layer.weight": "model-00001-of-00004.safetensors",
630
+ "mlm.base_model.model.model.transformer.blocks.5.attn_out.lora_A.default.weight": "model-00001-of-00004.safetensors",
631
+ "mlm.base_model.model.model.transformer.blocks.5.attn_out.lora_B.default.weight": "model-00001-of-00004.safetensors",
632
+ "mlm.base_model.model.model.transformer.blocks.5.ff_norm.weight": "model-00001-of-00004.safetensors",
633
+ "mlm.base_model.model.model.transformer.blocks.5.ff_out.base_layer.weight": "model-00001-of-00004.safetensors",
634
+ "mlm.base_model.model.model.transformer.blocks.5.ff_out.lora_A.default.weight": "model-00001-of-00004.safetensors",
635
+ "mlm.base_model.model.model.transformer.blocks.5.ff_out.lora_B.default.weight": "model-00001-of-00004.safetensors",
636
+ "mlm.base_model.model.model.transformer.blocks.5.ff_proj.base_layer.weight": "model-00001-of-00004.safetensors",
637
+ "mlm.base_model.model.model.transformer.blocks.5.ff_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
638
+ "mlm.base_model.model.model.transformer.blocks.5.ff_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
639
+ "mlm.base_model.model.model.transformer.blocks.5.k_proj.base_layer.weight": "model-00001-of-00004.safetensors",
640
+ "mlm.base_model.model.model.transformer.blocks.5.k_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
641
+ "mlm.base_model.model.model.transformer.blocks.5.k_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
642
+ "mlm.base_model.model.model.transformer.blocks.5.q_proj.base_layer.weight": "model-00001-of-00004.safetensors",
643
+ "mlm.base_model.model.model.transformer.blocks.5.q_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
644
+ "mlm.base_model.model.model.transformer.blocks.5.q_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
645
+ "mlm.base_model.model.model.transformer.blocks.5.up_proj.base_layer.weight": "model-00001-of-00004.safetensors",
646
+ "mlm.base_model.model.model.transformer.blocks.5.up_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
647
+ "mlm.base_model.model.model.transformer.blocks.5.up_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
648
+ "mlm.base_model.model.model.transformer.blocks.5.v_proj.base_layer.weight": "model-00001-of-00004.safetensors",
649
+ "mlm.base_model.model.model.transformer.blocks.5.v_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
650
+ "mlm.base_model.model.model.transformer.blocks.5.v_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
651
+ "mlm.base_model.model.model.transformer.blocks.6.attn_norm.weight": "model-00001-of-00004.safetensors",
652
+ "mlm.base_model.model.model.transformer.blocks.6.attn_out.base_layer.weight": "model-00001-of-00004.safetensors",
653
+ "mlm.base_model.model.model.transformer.blocks.6.attn_out.lora_A.default.weight": "model-00001-of-00004.safetensors",
654
+ "mlm.base_model.model.model.transformer.blocks.6.attn_out.lora_B.default.weight": "model-00001-of-00004.safetensors",
655
+ "mlm.base_model.model.model.transformer.blocks.6.ff_norm.weight": "model-00001-of-00004.safetensors",
656
+ "mlm.base_model.model.model.transformer.blocks.6.ff_out.base_layer.weight": "model-00001-of-00004.safetensors",
657
+ "mlm.base_model.model.model.transformer.blocks.6.ff_out.lora_A.default.weight": "model-00001-of-00004.safetensors",
658
+ "mlm.base_model.model.model.transformer.blocks.6.ff_out.lora_B.default.weight": "model-00001-of-00004.safetensors",
659
+ "mlm.base_model.model.model.transformer.blocks.6.ff_proj.base_layer.weight": "model-00001-of-00004.safetensors",
660
+ "mlm.base_model.model.model.transformer.blocks.6.ff_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
661
+ "mlm.base_model.model.model.transformer.blocks.6.ff_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
662
+ "mlm.base_model.model.model.transformer.blocks.6.k_proj.base_layer.weight": "model-00001-of-00004.safetensors",
663
+ "mlm.base_model.model.model.transformer.blocks.6.k_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
664
+ "mlm.base_model.model.model.transformer.blocks.6.k_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
665
+ "mlm.base_model.model.model.transformer.blocks.6.q_proj.base_layer.weight": "model-00001-of-00004.safetensors",
666
+ "mlm.base_model.model.model.transformer.blocks.6.q_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
667
+ "mlm.base_model.model.model.transformer.blocks.6.q_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
668
+ "mlm.base_model.model.model.transformer.blocks.6.up_proj.base_layer.weight": "model-00001-of-00004.safetensors",
669
+ "mlm.base_model.model.model.transformer.blocks.6.up_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
670
+ "mlm.base_model.model.model.transformer.blocks.6.up_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
671
+ "mlm.base_model.model.model.transformer.blocks.6.v_proj.base_layer.weight": "model-00001-of-00004.safetensors",
672
+ "mlm.base_model.model.model.transformer.blocks.6.v_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
673
+ "mlm.base_model.model.model.transformer.blocks.6.v_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
674
+ "mlm.base_model.model.model.transformer.blocks.7.attn_norm.weight": "model-00001-of-00004.safetensors",
675
+ "mlm.base_model.model.model.transformer.blocks.7.attn_out.base_layer.weight": "model-00001-of-00004.safetensors",
676
+ "mlm.base_model.model.model.transformer.blocks.7.attn_out.lora_A.default.weight": "model-00001-of-00004.safetensors",
677
+ "mlm.base_model.model.model.transformer.blocks.7.attn_out.lora_B.default.weight": "model-00001-of-00004.safetensors",
678
+ "mlm.base_model.model.model.transformer.blocks.7.ff_norm.weight": "model-00001-of-00004.safetensors",
679
+ "mlm.base_model.model.model.transformer.blocks.7.ff_out.base_layer.weight": "model-00001-of-00004.safetensors",
680
+ "mlm.base_model.model.model.transformer.blocks.7.ff_out.lora_A.default.weight": "model-00001-of-00004.safetensors",
681
+ "mlm.base_model.model.model.transformer.blocks.7.ff_out.lora_B.default.weight": "model-00001-of-00004.safetensors",
682
+ "mlm.base_model.model.model.transformer.blocks.7.ff_proj.base_layer.weight": "model-00001-of-00004.safetensors",
683
+ "mlm.base_model.model.model.transformer.blocks.7.ff_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
684
+ "mlm.base_model.model.model.transformer.blocks.7.ff_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
685
+ "mlm.base_model.model.model.transformer.blocks.7.k_proj.base_layer.weight": "model-00001-of-00004.safetensors",
686
+ "mlm.base_model.model.model.transformer.blocks.7.k_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
687
+ "mlm.base_model.model.model.transformer.blocks.7.k_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
688
+ "mlm.base_model.model.model.transformer.blocks.7.q_proj.base_layer.weight": "model-00001-of-00004.safetensors",
689
+ "mlm.base_model.model.model.transformer.blocks.7.q_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
690
+ "mlm.base_model.model.model.transformer.blocks.7.q_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
691
+ "mlm.base_model.model.model.transformer.blocks.7.up_proj.base_layer.weight": "model-00001-of-00004.safetensors",
692
+ "mlm.base_model.model.model.transformer.blocks.7.up_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
693
+ "mlm.base_model.model.model.transformer.blocks.7.up_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
694
+ "mlm.base_model.model.model.transformer.blocks.7.v_proj.base_layer.weight": "model-00001-of-00004.safetensors",
695
+ "mlm.base_model.model.model.transformer.blocks.7.v_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
696
+ "mlm.base_model.model.model.transformer.blocks.7.v_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
697
+ "mlm.base_model.model.model.transformer.blocks.8.attn_norm.weight": "model-00001-of-00004.safetensors",
698
+ "mlm.base_model.model.model.transformer.blocks.8.attn_out.base_layer.weight": "model-00001-of-00004.safetensors",
699
+ "mlm.base_model.model.model.transformer.blocks.8.attn_out.lora_A.default.weight": "model-00001-of-00004.safetensors",
700
+ "mlm.base_model.model.model.transformer.blocks.8.attn_out.lora_B.default.weight": "model-00001-of-00004.safetensors",
701
+ "mlm.base_model.model.model.transformer.blocks.8.ff_norm.weight": "model-00001-of-00004.safetensors",
702
+ "mlm.base_model.model.model.transformer.blocks.8.ff_out.base_layer.weight": "model-00001-of-00004.safetensors",
703
+ "mlm.base_model.model.model.transformer.blocks.8.ff_out.lora_A.default.weight": "model-00001-of-00004.safetensors",
704
+ "mlm.base_model.model.model.transformer.blocks.8.ff_out.lora_B.default.weight": "model-00001-of-00004.safetensors",
705
+ "mlm.base_model.model.model.transformer.blocks.8.ff_proj.base_layer.weight": "model-00001-of-00004.safetensors",
706
+ "mlm.base_model.model.model.transformer.blocks.8.ff_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
707
+ "mlm.base_model.model.model.transformer.blocks.8.ff_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
708
+ "mlm.base_model.model.model.transformer.blocks.8.k_proj.base_layer.weight": "model-00001-of-00004.safetensors",
709
+ "mlm.base_model.model.model.transformer.blocks.8.k_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
710
+ "mlm.base_model.model.model.transformer.blocks.8.k_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
711
+ "mlm.base_model.model.model.transformer.blocks.8.q_proj.base_layer.weight": "model-00001-of-00004.safetensors",
712
+ "mlm.base_model.model.model.transformer.blocks.8.q_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
713
+ "mlm.base_model.model.model.transformer.blocks.8.q_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
714
+ "mlm.base_model.model.model.transformer.blocks.8.up_proj.base_layer.weight": "model-00001-of-00004.safetensors",
715
+ "mlm.base_model.model.model.transformer.blocks.8.up_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
716
+ "mlm.base_model.model.model.transformer.blocks.8.up_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
717
+ "mlm.base_model.model.model.transformer.blocks.8.v_proj.base_layer.weight": "model-00001-of-00004.safetensors",
718
+ "mlm.base_model.model.model.transformer.blocks.8.v_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
719
+ "mlm.base_model.model.model.transformer.blocks.8.v_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
720
+ "mlm.base_model.model.model.transformer.blocks.9.attn_norm.weight": "model-00001-of-00004.safetensors",
721
+ "mlm.base_model.model.model.transformer.blocks.9.attn_out.base_layer.weight": "model-00001-of-00004.safetensors",
722
+ "mlm.base_model.model.model.transformer.blocks.9.attn_out.lora_A.default.weight": "model-00001-of-00004.safetensors",
723
+ "mlm.base_model.model.model.transformer.blocks.9.attn_out.lora_B.default.weight": "model-00001-of-00004.safetensors",
724
+ "mlm.base_model.model.model.transformer.blocks.9.ff_norm.weight": "model-00001-of-00004.safetensors",
725
+ "mlm.base_model.model.model.transformer.blocks.9.ff_out.base_layer.weight": "model-00001-of-00004.safetensors",
726
+ "mlm.base_model.model.model.transformer.blocks.9.ff_out.lora_A.default.weight": "model-00001-of-00004.safetensors",
727
+ "mlm.base_model.model.model.transformer.blocks.9.ff_out.lora_B.default.weight": "model-00001-of-00004.safetensors",
728
+ "mlm.base_model.model.model.transformer.blocks.9.ff_proj.base_layer.weight": "model-00002-of-00004.safetensors",
729
+ "mlm.base_model.model.model.transformer.blocks.9.ff_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
730
+ "mlm.base_model.model.model.transformer.blocks.9.ff_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
731
+ "mlm.base_model.model.model.transformer.blocks.9.k_proj.base_layer.weight": "model-00001-of-00004.safetensors",
732
+ "mlm.base_model.model.model.transformer.blocks.9.k_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
733
+ "mlm.base_model.model.model.transformer.blocks.9.k_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
734
+ "mlm.base_model.model.model.transformer.blocks.9.q_proj.base_layer.weight": "model-00001-of-00004.safetensors",
735
+ "mlm.base_model.model.model.transformer.blocks.9.q_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
736
+ "mlm.base_model.model.model.transformer.blocks.9.q_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
737
+ "mlm.base_model.model.model.transformer.blocks.9.up_proj.base_layer.weight": "model-00002-of-00004.safetensors",
738
+ "mlm.base_model.model.model.transformer.blocks.9.up_proj.lora_A.default.weight": "model-00002-of-00004.safetensors",
739
+ "mlm.base_model.model.model.transformer.blocks.9.up_proj.lora_B.default.weight": "model-00002-of-00004.safetensors",
740
+ "mlm.base_model.model.model.transformer.blocks.9.v_proj.base_layer.weight": "model-00001-of-00004.safetensors",
741
+ "mlm.base_model.model.model.transformer.blocks.9.v_proj.lora_A.default.weight": "model-00001-of-00004.safetensors",
742
+ "mlm.base_model.model.model.transformer.blocks.9.v_proj.lora_B.default.weight": "model-00001-of-00004.safetensors",
743
+ "mlm.base_model.model.model.transformer.ff_out.base_layer.weight": "model-00004-of-00004.safetensors",
744
+ "mlm.base_model.model.model.transformer.ff_out.lora_A.default.weight": "model-00004-of-00004.safetensors",
745
+ "mlm.base_model.model.model.transformer.ff_out.lora_B.default.weight": "model-00004-of-00004.safetensors",
746
+ "mlm.base_model.model.model.transformer.ln_f.weight": "model-00001-of-00004.safetensors",
747
+ "mlm.base_model.model.model.transformer.wte.base_layer.weight": "model-00001-of-00004.safetensors",
748
+ "mlm.base_model.model.model.transformer.wte.lora_embedding_A.default": "model-00001-of-00004.safetensors",
749
+ "mlm.base_model.model.model.transformer.wte.lora_embedding_B.default": "model-00001-of-00004.safetensors"
750
+ }
751
+ }
special_tokens_map.json ADDED
@@ -0,0 +1,46 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<|mdm_mask|>",
4
+ "<role>",
5
+ "</role>",
6
+ "<|arithmetic_start|>",
7
+ "<|arithmetic_end|>",
8
+ "<|number_start|>",
9
+ "<|number_end|>"
10
+ ],
11
+ "bos_token": {
12
+ "content": "<|startoftext|>",
13
+ "lstrip": false,
14
+ "normalized": false,
15
+ "rstrip": false,
16
+ "single_word": false
17
+ },
18
+ "cls_token": {
19
+ "content": "[CLS]",
20
+ "lstrip": false,
21
+ "normalized": false,
22
+ "rstrip": false,
23
+ "single_word": false
24
+ },
25
+ "eos_token": {
26
+ "content": "<|endoftext|>",
27
+ "lstrip": false,
28
+ "normalized": false,
29
+ "rstrip": false,
30
+ "single_word": false
31
+ },
32
+ "mask_token": {
33
+ "content": "<|mdm_mask|>",
34
+ "lstrip": false,
35
+ "normalized": false,
36
+ "rstrip": false,
37
+ "single_word": false
38
+ },
39
+ "pad_token": {
40
+ "content": "<|endoftext|>",
41
+ "lstrip": false,
42
+ "normalized": false,
43
+ "rstrip": false,
44
+ "single_word": false
45
+ }
46
+ }
tokenizer.json ADDED
@@ -0,0 +1,217 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "version": "1.0",
3
+ "truncation": null,
4
+ "padding": null,
5
+ "added_tokens": [
6
+ {
7
+ "id": 75,
8
+ "content": "<|startoftext|>",
9
+ "single_word": false,
10
+ "lstrip": false,
11
+ "rstrip": false,
12
+ "normalized": false,
13
+ "special": true
14
+ },
15
+ {
16
+ "id": 76,
17
+ "content": "<|endoftext|>",
18
+ "single_word": false,
19
+ "lstrip": false,
20
+ "rstrip": false,
21
+ "normalized": false,
22
+ "special": true
23
+ },
24
+ {
25
+ "id": 77,
26
+ "content": "[CLS]",
27
+ "single_word": false,
28
+ "lstrip": false,
29
+ "rstrip": false,
30
+ "normalized": false,
31
+ "special": true
32
+ },
33
+ {
34
+ "id": 78,
35
+ "content": "<|mdm_mask|>",
36
+ "single_word": false,
37
+ "lstrip": false,
38
+ "rstrip": false,
39
+ "normalized": false,
40
+ "special": true
41
+ },
42
+ {
43
+ "id": 79,
44
+ "content": "<role>",
45
+ "single_word": false,
46
+ "lstrip": false,
47
+ "rstrip": false,
48
+ "normalized": false,
49
+ "special": true
50
+ },
51
+ {
52
+ "id": 80,
53
+ "content": "</role>",
54
+ "single_word": false,
55
+ "lstrip": false,
56
+ "rstrip": false,
57
+ "normalized": false,
58
+ "special": true
59
+ },
60
+ {
61
+ "id": 81,
62
+ "content": "<|arithmetic_start|>",
63
+ "single_word": false,
64
+ "lstrip": false,
65
+ "rstrip": false,
66
+ "normalized": false,
67
+ "special": true
68
+ },
69
+ {
70
+ "id": 82,
71
+ "content": "<|arithmetic_end|>",
72
+ "single_word": false,
73
+ "lstrip": false,
74
+ "rstrip": false,
75
+ "normalized": false,
76
+ "special": true
77
+ },
78
+ {
79
+ "id": 83,
80
+ "content": "<|number_start|>",
81
+ "single_word": false,
82
+ "lstrip": false,
83
+ "rstrip": false,
84
+ "normalized": false,
85
+ "special": true
86
+ },
87
+ {
88
+ "id": 84,
89
+ "content": "<|number_end|>",
90
+ "single_word": false,
91
+ "lstrip": false,
92
+ "rstrip": false,
93
+ "normalized": false,
94
+ "special": true
95
+ }
96
+ ],
97
+ "normalizer": null,
98
+ "pre_tokenizer": {
99
+ "type": "Sequence",
100
+ "pretokenizers": [
101
+ {
102
+ "type": "Split",
103
+ "pattern": {
104
+ "Regex": "'(?i:[sdmt]|ll|ve|re)|[^\\r\\n\\p{L}\\p{N}]?+\\p{L}+|\\p{N}| ?[^\\s\\p{L}\\p{N}]++[\\r\\n]*|\\s*[\\r\\n]|\\s+(?!\\S)|\\s+"
105
+ },
106
+ "behavior": "Isolated",
107
+ "invert": false
108
+ },
109
+ {
110
+ "type": "ByteLevel",
111
+ "add_prefix_space": false,
112
+ "trim_offsets": true,
113
+ "use_regex": false
114
+ }
115
+ ]
116
+ },
117
+ "post_processor": {
118
+ "type": "ByteLevel",
119
+ "add_prefix_space": true,
120
+ "trim_offsets": false,
121
+ "use_regex": true
122
+ },
123
+ "decoder": {
124
+ "type": "ByteLevel",
125
+ "add_prefix_space": true,
126
+ "trim_offsets": true,
127
+ "use_regex": true
128
+ },
129
+ "model": {
130
+ "type": "BPE",
131
+ "dropout": null,
132
+ "unk_token": null,
133
+ "continuing_subword_prefix": null,
134
+ "end_of_word_suffix": null,
135
+ "fuse_unk": false,
136
+ "byte_fallback": false,
137
+ "ignore_merges": false,
138
+ "vocab": {
139
+ "!": 0,
140
+ "*": 1,
141
+ "+": 2,
142
+ ",": 3,
143
+ "-": 4,
144
+ ".": 5,
145
+ "/": 6,
146
+ "0": 7,
147
+ "1": 8,
148
+ "2": 9,
149
+ "3": 10,
150
+ "4": 11,
151
+ "5": 12,
152
+ "6": 13,
153
+ "7": 14,
154
+ "8": 15,
155
+ "9": 16,
156
+ ":": 17,
157
+ ";": 18,
158
+ "=": 19,
159
+ "?": 20,
160
+ "A": 21,
161
+ "B": 22,
162
+ "C": 23,
163
+ "D": 24,
164
+ "E": 25,
165
+ "F": 26,
166
+ "G": 27,
167
+ "H": 28,
168
+ "I": 29,
169
+ "J": 30,
170
+ "K": 31,
171
+ "L": 32,
172
+ "M": 33,
173
+ "N": 34,
174
+ "O": 35,
175
+ "P": 36,
176
+ "Q": 37,
177
+ "R": 38,
178
+ "S": 39,
179
+ "T": 40,
180
+ "U": 41,
181
+ "V": 42,
182
+ "W": 43,
183
+ "X": 44,
184
+ "Y": 45,
185
+ "Z": 46,
186
+ "a": 47,
187
+ "b": 48,
188
+ "c": 49,
189
+ "d": 50,
190
+ "e": 51,
191
+ "f": 52,
192
+ "g": 53,
193
+ "h": 54,
194
+ "i": 55,
195
+ "j": 56,
196
+ "k": 57,
197
+ "l": 58,
198
+ "m": 59,
199
+ "n": 60,
200
+ "o": 61,
201
+ "p": 62,
202
+ "q": 63,
203
+ "r": 64,
204
+ "s": 65,
205
+ "t": 66,
206
+ "u": 67,
207
+ "v": 68,
208
+ "w": 69,
209
+ "x": 70,
210
+ "y": 71,
211
+ "z": 72,
212
+ "|": 73,
213
+ "Ċ": 74
214
+ },
215
+ "merges": []
216
+ }
217
+ }
tokenizer_config.json ADDED
@@ -0,0 +1,113 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": false,
3
+ "add_eos_token": false,
4
+ "added_tokens_decoder": {
5
+ "75": {
6
+ "content": "<|startoftext|>",
7
+ "lstrip": false,
8
+ "normalized": false,
9
+ "rstrip": false,
10
+ "single_word": false,
11
+ "special": true
12
+ },
13
+ "76": {
14
+ "content": "<|endoftext|>",
15
+ "lstrip": false,
16
+ "normalized": false,
17
+ "rstrip": false,
18
+ "single_word": false,
19
+ "special": true
20
+ },
21
+ "77": {
22
+ "content": "[CLS]",
23
+ "lstrip": false,
24
+ "normalized": false,
25
+ "rstrip": false,
26
+ "single_word": false,
27
+ "special": true
28
+ },
29
+ "78": {
30
+ "content": "<|mdm_mask|>",
31
+ "lstrip": false,
32
+ "normalized": false,
33
+ "rstrip": false,
34
+ "single_word": false,
35
+ "special": true
36
+ },
37
+ "79": {
38
+ "content": "<role>",
39
+ "lstrip": false,
40
+ "normalized": false,
41
+ "rstrip": false,
42
+ "single_word": false,
43
+ "special": true
44
+ },
45
+ "80": {
46
+ "content": "</role>",
47
+ "lstrip": false,
48
+ "normalized": false,
49
+ "rstrip": false,
50
+ "single_word": false,
51
+ "special": true
52
+ },
53
+ "81": {
54
+ "content": "<|arithmetic_start|>",
55
+ "lstrip": false,
56
+ "normalized": false,
57
+ "rstrip": false,
58
+ "single_word": false,
59
+ "special": true
60
+ },
61
+ "82": {
62
+ "content": "<|arithmetic_end|>",
63
+ "lstrip": false,
64
+ "normalized": false,
65
+ "rstrip": false,
66
+ "single_word": false,
67
+ "special": true
68
+ },
69
+ "83": {
70
+ "content": "<|number_start|>",
71
+ "lstrip": false,
72
+ "normalized": false,
73
+ "rstrip": false,
74
+ "single_word": false,
75
+ "special": true
76
+ },
77
+ "84": {
78
+ "content": "<|number_end|>",
79
+ "lstrip": false,
80
+ "normalized": false,
81
+ "rstrip": false,
82
+ "single_word": false,
83
+ "special": true
84
+ }
85
+ },
86
+ "additional_special_tokens": [
87
+ "<|mdm_mask|>",
88
+ "<role>",
89
+ "</role>",
90
+ "<|arithmetic_start|>",
91
+ "<|arithmetic_end|>",
92
+ "<|number_start|>",
93
+ "<|number_end|>"
94
+ ],
95
+ "bos_token": "<|startoftext|>",
96
+ "clean_up_tokenization_spaces": false,
97
+ "cls_token": "[CLS]",
98
+ "eos_token": "<|endoftext|>",
99
+ "extra_special_tokens": {},
100
+ "fast_tokenizer": true,
101
+ "gmask_token": "[gMASK]",
102
+ "mask_token": "<|mdm_mask|>",
103
+ "merges_file": null,
104
+ "model_input_names": [
105
+ "input_ids",
106
+ "attention_mask"
107
+ ],
108
+ "model_max_length": 1000000000000000019884624838656,
109
+ "pad_token": "<|endoftext|>",
110
+ "padding_side": "right",
111
+ "tokenizer_class": "PreTrainedTokenizerFast",
112
+ "trust_remote_code": true
113
+ }