End of training
Browse files- README.md +169 -0
- adapter_model.bin +3 -0
README.md
ADDED
@@ -0,0 +1,169 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: llama3
|
3 |
+
library_name: peft
|
4 |
+
tags:
|
5 |
+
- axolotl
|
6 |
+
- generated_from_trainer
|
7 |
+
base_model: meta-llama/Meta-Llama-3-8B
|
8 |
+
model-index:
|
9 |
+
- name: isafpr-llama3-lora-templatefree
|
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 |
+
base_model: meta-llama/Meta-Llama-3-8B
|
22 |
+
model_type: LlamaForCausalLM
|
23 |
+
tokenizer_type: AutoTokenizer
|
24 |
+
|
25 |
+
load_in_8bit: false
|
26 |
+
load_in_4bit: true
|
27 |
+
strict: false
|
28 |
+
|
29 |
+
data_seed: 42
|
30 |
+
seed: 42
|
31 |
+
|
32 |
+
datasets:
|
33 |
+
- path: data/templatefree_isaf_press_releases_ft_train.jsonl
|
34 |
+
type: input_output
|
35 |
+
dataset_prepared_path:
|
36 |
+
val_set_size: 0.1
|
37 |
+
output_dir: ./outputs/llama3/lora-out-templatefree
|
38 |
+
hub_model_id: strickvl/isafpr-llama3-lora-templatefree
|
39 |
+
|
40 |
+
sequence_len: 1024
|
41 |
+
sample_packing: true
|
42 |
+
eval_sample_packing: false
|
43 |
+
pad_to_sequence_len: true
|
44 |
+
|
45 |
+
adapter: lora
|
46 |
+
lora_model_dir:
|
47 |
+
lora_r: 32
|
48 |
+
lora_alpha: 16
|
49 |
+
lora_dropout: 0.05
|
50 |
+
lora_target_linear: true
|
51 |
+
lora_fan_in_fan_out:
|
52 |
+
lora_modules_to_save:
|
53 |
+
- embed_tokens
|
54 |
+
- lm_head
|
55 |
+
|
56 |
+
wandb_project: isaf_pr_ft
|
57 |
+
wandb_entity: strickvl
|
58 |
+
wandb_watch:
|
59 |
+
wandb_name:
|
60 |
+
wandb_log_model:
|
61 |
+
|
62 |
+
gradient_accumulation_steps: 4
|
63 |
+
micro_batch_size: 2
|
64 |
+
num_epochs: 4
|
65 |
+
optimizer: adamw_bnb_8bit
|
66 |
+
lr_scheduler: cosine
|
67 |
+
learning_rate: 0.0002
|
68 |
+
|
69 |
+
train_on_inputs: false
|
70 |
+
group_by_length: false
|
71 |
+
bf16: auto
|
72 |
+
fp16:
|
73 |
+
tf32: false
|
74 |
+
|
75 |
+
gradient_checkpointing: true
|
76 |
+
early_stopping_patience:
|
77 |
+
resume_from_checkpoint:
|
78 |
+
local_rank:
|
79 |
+
logging_steps: 1
|
80 |
+
xformers_attention:
|
81 |
+
flash_attention: true
|
82 |
+
s2_attention:
|
83 |
+
|
84 |
+
warmup_steps: 10
|
85 |
+
evals_per_epoch: 4
|
86 |
+
eval_table_size:
|
87 |
+
eval_max_new_tokens: 128
|
88 |
+
saves_per_epoch: 1
|
89 |
+
debug:
|
90 |
+
deepspeed:
|
91 |
+
weight_decay: 0.0
|
92 |
+
fsdp:
|
93 |
+
fsdp_config:
|
94 |
+
special_tokens:
|
95 |
+
bos_token: "<s>"
|
96 |
+
eos_token: "</s>"
|
97 |
+
pad_token: <|end_of_text|>
|
98 |
+
|
99 |
+
```
|
100 |
+
|
101 |
+
</details><br>
|
102 |
+
|
103 |
+
# isafpr-llama3-lora-templatefree
|
104 |
+
|
105 |
+
This model is a fine-tuned version of [meta-llama/Meta-Llama-3-8B](https://huggingface.co/meta-llama/Meta-Llama-3-8B) on the None dataset.
|
106 |
+
It achieves the following results on the evaluation set:
|
107 |
+
- Loss: 0.0428
|
108 |
+
|
109 |
+
## Model description
|
110 |
+
|
111 |
+
More information needed
|
112 |
+
|
113 |
+
## Intended uses & limitations
|
114 |
+
|
115 |
+
More information needed
|
116 |
+
|
117 |
+
## Training and evaluation data
|
118 |
+
|
119 |
+
More information needed
|
120 |
+
|
121 |
+
## Training procedure
|
122 |
+
|
123 |
+
### Training hyperparameters
|
124 |
+
|
125 |
+
The following hyperparameters were used during training:
|
126 |
+
- learning_rate: 0.0002
|
127 |
+
- train_batch_size: 2
|
128 |
+
- eval_batch_size: 2
|
129 |
+
- seed: 42
|
130 |
+
- distributed_type: multi-GPU
|
131 |
+
- num_devices: 2
|
132 |
+
- gradient_accumulation_steps: 4
|
133 |
+
- total_train_batch_size: 16
|
134 |
+
- total_eval_batch_size: 4
|
135 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
136 |
+
- lr_scheduler_type: cosine
|
137 |
+
- lr_scheduler_warmup_steps: 10
|
138 |
+
- num_epochs: 4
|
139 |
+
|
140 |
+
### Training results
|
141 |
+
|
142 |
+
| Training Loss | Epoch | Step | Validation Loss |
|
143 |
+
|:-------------:|:------:|:----:|:---------------:|
|
144 |
+
| 2.1433 | 0.0071 | 1 | 2.1450 |
|
145 |
+
| 0.0712 | 0.25 | 35 | 0.0669 |
|
146 |
+
| 0.0549 | 0.5 | 70 | 0.0517 |
|
147 |
+
| 0.0585 | 0.75 | 105 | 0.0479 |
|
148 |
+
| 0.0452 | 1.0 | 140 | 0.0482 |
|
149 |
+
| 0.0244 | 1.2339 | 175 | 0.0473 |
|
150 |
+
| 0.0287 | 1.4839 | 210 | 0.0447 |
|
151 |
+
| 0.017 | 1.7339 | 245 | 0.0417 |
|
152 |
+
| 0.0107 | 1.9839 | 280 | 0.0408 |
|
153 |
+
| 0.0151 | 2.2143 | 315 | 0.0414 |
|
154 |
+
| 0.0134 | 2.4643 | 350 | 0.0415 |
|
155 |
+
| 0.0067 | 2.7143 | 385 | 0.0407 |
|
156 |
+
| 0.0089 | 2.9643 | 420 | 0.0399 |
|
157 |
+
| 0.0092 | 3.1929 | 455 | 0.0421 |
|
158 |
+
| 0.007 | 3.4429 | 490 | 0.0429 |
|
159 |
+
| 0.0065 | 3.6929 | 525 | 0.0428 |
|
160 |
+
| 0.0125 | 3.9429 | 560 | 0.0428 |
|
161 |
+
|
162 |
+
|
163 |
+
### Framework versions
|
164 |
+
|
165 |
+
- PEFT 0.11.1
|
166 |
+
- Transformers 4.41.1
|
167 |
+
- Pytorch 2.3.0+cu121
|
168 |
+
- Datasets 2.19.1
|
169 |
+
- Tokenizers 0.19.1
|
adapter_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:51e3730eea2e989711d0b3e0245a267e1634432298be47f5d580b749e357b787
|
3 |
+
size 4538466622
|