leafspark commited on
Commit
39b2f86
·
verified ·
1 Parent(s): 58d72bd

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +95 -3
README.md CHANGED
@@ -1,3 +1,95 @@
1
- ---
2
- license: llama3
3
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: llama3
3
+ language:
4
+ - en
5
+ library_name: transformers
6
+ pipeline_tag: text-generation
7
+ datasets:
8
+ - cognitivecomputations/WizardLM_alpaca_evol_instruct_70k_unfiltered
9
+ tags:
10
+ - finetune
11
+ - llama
12
+ ---
13
+ # WizardHermes-2-Pro-Llama-3-8b
14
+ Finetuned from Meta-Llama-3-8b, 0.55 epoches
15
+
16
+ Merged with Hermes-2-Pro-Llama-3-8b
17
+
18
+ Eval (0.0 temp) ([https://inflection.ai/assets/MMLU-Examples.pdf](https://inflection.ai/assets/MMLU-Examples.pdf)):
19
+ ```
20
+ Hermes 2 Pro LoRA (merged) (checkpoint 1): 7/12
21
+ Hermes 2 Pro LoRA (merged) (checkpoint 2): 9/12
22
+ Hermes 2 Pro Llama 3 8b: 8/12
23
+ ```
24
+
25
+ Trained using MonsterAPI
26
+
27
+ LoRA training parameters:
28
+ ```
29
+ {
30
+ "deployment_name": "wizardlm-llama-3-8b",
31
+ "pretrainedmodel_config": {
32
+ "model_path": "meta-llama/Meta-Llama-3-8B",
33
+ "other_model_info": null,
34
+ "resume_checkpoint_path": "",
35
+ "use_lora": true,
36
+ "lora_r": 8,
37
+ "lora_alpha": 16,
38
+ "lora_dropout": 0,
39
+ "lora_bias": "none",
40
+ "use_quantization": false,
41
+ "use_gradient_checkpointing": false,
42
+ "parallelization": "nmp"
43
+ },
44
+ "data_config": {
45
+ "data_path": "cognitivecomputations/WizardLM_alpaca_evol_instruct_70k_unfiltered",
46
+ "data_subset": "default",
47
+ "data_source_type": "hub_link",
48
+ "cutoff_len": 4096,
49
+ "data_split_config": {
50
+ "train": 0.9,
51
+ "validation": 0.1
52
+ },
53
+ "prevalidated": true,
54
+ "concat_config": {
55
+ "0": {
56
+ "text": "Below is an instruction that describes a task. Write a response that appropriately completes the request. \n\n\n###Instruction:",
57
+ "column": false
58
+ },
59
+ "1": {
60
+ "text": "instruction",
61
+ "column": true
62
+ },
63
+ "2": {
64
+ "text": "\n\n\n###Response:",
65
+ "column": false
66
+ },
67
+ "3": {
68
+ "text": "output",
69
+ "column": true
70
+ }
71
+ }
72
+ },
73
+ "training_config": {
74
+ "early_stopping_patience": 5,
75
+ "num_train_epochs": 5,
76
+ "gradient_accumulation_steps": 1,
77
+ "warmup_steps": 300,
78
+ "learning_rate": 0.0005,
79
+ "lr_scheduler_type": "reduce_lr_on_plateau",
80
+ "group_by_length": false,
81
+ "use_hugging_face": false
82
+ },
83
+ "logging_config": {
84
+ "use_wandb": false,
85
+ "wandb_username": "",
86
+ "wandb_login_key": "",
87
+ "wandb_project": "",
88
+ "wandb_run_name": ""
89
+ },
90
+ "accessorytasks_config": {
91
+ "run_eval_report": false,
92
+ "run_quantize_merge": false
93
+ }
94
+ }
95
+ ```