shalini03 commited on
Commit
80c5c72
1 Parent(s): edd25f5

End of training

Browse files
Files changed (2) hide show
  1. README.md +154 -0
  2. adapter_model.bin +3 -0
README.md ADDED
@@ -0,0 +1,154 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ library_name: peft
4
+ tags:
5
+ - axolotl
6
+ - generated_from_trainer
7
+ base_model: TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T
8
+ model-index:
9
+ - name: tinyllama-1.1B_alpaca_2k_lora
10
+ results: []
11
+ ---
12
+
13
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
14
+ should probably proofread and complete it, then remove this comment. -->
15
+
16
+ [<img src="https://raw.githubusercontent.com/OpenAccess-AI-Collective/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/OpenAccess-AI-Collective/axolotl)
17
+ <details><summary>See axolotl config</summary>
18
+
19
+ axolotl version: `0.4.1`
20
+ ```yaml
21
+ # Upload the final model to Huggingface
22
+ hub_model_id: shalini03/tinyllama-1.1B_alpaca_2k_lora
23
+
24
+ # Store the training logs in weights and biases
25
+ #wandb_entity: shalini_03
26
+ #wandb_project: ft_tinyllama-1.1B_alpaca_2k_lora
27
+
28
+ # The rest of this config stays the same:
29
+ base_model: TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T
30
+ model_type: LlamaForCausalLM
31
+ tokenizer_type: LlamaTokenizer
32
+
33
+ load_in_8bit: true
34
+ load_in_4bit: false
35
+ strict: false
36
+
37
+ datasets:
38
+ - path: mhenrichsen/alpaca_2k_test
39
+ type: alpaca
40
+ dataset_prepared_path:
41
+ val_set_size: 0.05
42
+ output_dir: ./outputs/lora-out
43
+
44
+ sequence_len: 4096
45
+ sample_packing: true
46
+ eval_sample_packing: false
47
+ pad_to_sequence_len: true
48
+
49
+ adapter: lora
50
+ lora_model_dir:
51
+ lora_r: 32
52
+ lora_alpha: 16
53
+ lora_dropout: 0.05
54
+ lora_target_linear: true
55
+ lora_fan_in_fan_out:
56
+
57
+ gradient_accumulation_steps: 4
58
+ micro_batch_size: 2
59
+ num_epochs: 4
60
+ optimizer: adamw_bnb_8bit
61
+ lr_scheduler: cosine
62
+ learning_rate: 0.0002
63
+
64
+ train_on_inputs: false
65
+ group_by_length: false
66
+ bf16: auto
67
+ fp16:
68
+ tf32: false
69
+
70
+ gradient_checkpointing: true
71
+ early_stopping_patience:
72
+ resume_from_checkpoint:
73
+ local_rank:
74
+ logging_steps: 1
75
+ xformers_attention:
76
+ flash_attention: true
77
+
78
+ warmup_steps: 10
79
+ evals_per_epoch: 4
80
+ saves_per_epoch: 1
81
+ debug:
82
+ deepspeed:
83
+ weight_decay: 0.0
84
+ fsdp:
85
+ fsdp_config:
86
+ special_tokens:
87
+ ```
88
+
89
+ </details><br>
90
+
91
+ # tinyllama-1.1B_alpaca_2k_lora
92
+
93
+ This model is a fine-tuned version of [TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T](https://huggingface.co/TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T) on the None dataset.
94
+ It achieves the following results on the evaluation set:
95
+ - Loss: 1.2111
96
+
97
+ ## Model description
98
+
99
+ More information needed
100
+
101
+ ## Intended uses & limitations
102
+
103
+ More information needed
104
+
105
+ ## Training and evaluation data
106
+
107
+ More information needed
108
+
109
+ ## Training procedure
110
+
111
+ ### Training hyperparameters
112
+
113
+ The following hyperparameters were used during training:
114
+ - learning_rate: 0.0002
115
+ - train_batch_size: 2
116
+ - eval_batch_size: 2
117
+ - seed: 42
118
+ - gradient_accumulation_steps: 4
119
+ - total_train_batch_size: 8
120
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
121
+ - lr_scheduler_type: cosine
122
+ - lr_scheduler_warmup_steps: 10
123
+ - num_epochs: 4
124
+
125
+ ### Training results
126
+
127
+ | Training Loss | Epoch | Step | Validation Loss |
128
+ |:-------------:|:------:|:----:|:---------------:|
129
+ | 1.4615 | 0.0816 | 1 | 1.4899 |
130
+ | 1.3851 | 0.2449 | 3 | 1.4869 |
131
+ | 1.3658 | 0.4898 | 6 | 1.4376 |
132
+ | 1.2683 | 0.7347 | 9 | 1.3399 |
133
+ | 1.2259 | 0.9796 | 12 | 1.2956 |
134
+ | 1.2523 | 1.1633 | 15 | 1.2787 |
135
+ | 1.2271 | 1.4082 | 18 | 1.2527 |
136
+ | 1.1348 | 1.6531 | 21 | 1.2336 |
137
+ | 1.2694 | 1.8980 | 24 | 1.2286 |
138
+ | 1.1484 | 2.0816 | 27 | 1.2224 |
139
+ | 1.1527 | 2.3265 | 30 | 1.2214 |
140
+ | 1.1937 | 2.5714 | 33 | 1.2187 |
141
+ | 1.1121 | 2.8163 | 36 | 1.2150 |
142
+ | 1.1517 | 3.0612 | 39 | 1.2147 |
143
+ | 1.1888 | 3.2449 | 42 | 1.2107 |
144
+ | 1.1002 | 3.4898 | 45 | 1.2122 |
145
+ | 1.1884 | 3.7347 | 48 | 1.2111 |
146
+
147
+
148
+ ### Framework versions
149
+
150
+ - PEFT 0.11.1
151
+ - Transformers 4.41.1
152
+ - Pytorch 2.1.2+cu118
153
+ - Datasets 2.19.1
154
+ - Tokenizers 0.19.1
adapter_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5f05fd4e3f5bd62fdb54b68694aba59a2dfdf5e83f9a5a5cdc6fbda5de1ca650
3
+ size 101036698