End of training
Browse files- README.md +147 -0
- adapter_model.bin +3 -0
README.md
ADDED
@@ -0,0 +1,147 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
library_name: peft
|
3 |
+
license: other
|
4 |
+
base_model: unsloth/Qwen2.5-3B-Instruct
|
5 |
+
tags:
|
6 |
+
- axolotl
|
7 |
+
- generated_from_trainer
|
8 |
+
model-index:
|
9 |
+
- name: d39c2c44-497a-4263-b4dd-0b4ec14c1b5c
|
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/axolotl-ai-cloud/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/axolotl-ai-cloud/axolotl)
|
17 |
+
<details><summary>See axolotl config</summary>
|
18 |
+
|
19 |
+
axolotl version: `0.5.2`
|
20 |
+
```yaml
|
21 |
+
adapter: lora
|
22 |
+
base_model: unsloth/Qwen2.5-3B-Instruct
|
23 |
+
bf16: auto
|
24 |
+
chat_template: llama3
|
25 |
+
dataset_prepared_path: null
|
26 |
+
datasets:
|
27 |
+
- data_files:
|
28 |
+
- f900285c16573485_train_data.json
|
29 |
+
ds_type: json
|
30 |
+
field: labels_str
|
31 |
+
path: /workspace/input_data/f900285c16573485_train_data.json
|
32 |
+
type: completion
|
33 |
+
debug: null
|
34 |
+
deepspeed: null
|
35 |
+
early_stopping_patience: 1
|
36 |
+
eval_max_new_tokens: 128
|
37 |
+
eval_steps: 25
|
38 |
+
eval_table_size: null
|
39 |
+
flash_attention: false
|
40 |
+
fp16: false
|
41 |
+
fsdp: null
|
42 |
+
fsdp_config: null
|
43 |
+
gradient_accumulation_steps: 16
|
44 |
+
gradient_checkpointing: true
|
45 |
+
group_by_length: true
|
46 |
+
hub_model_id: vdos/d39c2c44-497a-4263-b4dd-0b4ec14c1b5c
|
47 |
+
hub_repo: null
|
48 |
+
hub_strategy: checkpoint
|
49 |
+
hub_token: null
|
50 |
+
learning_rate: 0.0001
|
51 |
+
load_in_4bit: false
|
52 |
+
load_in_8bit: false
|
53 |
+
local_rank: null
|
54 |
+
logging_steps: 1
|
55 |
+
lora_alpha: 64
|
56 |
+
lora_dropout: 0.05
|
57 |
+
lora_fan_in_fan_out: null
|
58 |
+
lora_model_dir: null
|
59 |
+
lora_r: 32
|
60 |
+
lora_target_linear: true
|
61 |
+
lr_scheduler: cosine
|
62 |
+
max_steps: 50
|
63 |
+
micro_batch_size: 2
|
64 |
+
mlflow_experiment_name: /tmp/f900285c16573485_train_data.json
|
65 |
+
model_type: AutoModelForCausalLM
|
66 |
+
num_epochs: 3
|
67 |
+
optimizer: adamw_torch
|
68 |
+
output_dir: miner_id_24
|
69 |
+
pad_to_sequence_len: true
|
70 |
+
resume_from_checkpoint: null
|
71 |
+
s2_attention: null
|
72 |
+
sample_packing: false
|
73 |
+
save_steps: 25
|
74 |
+
sequence_len: 2048
|
75 |
+
strict: false
|
76 |
+
tf32: false
|
77 |
+
tokenizer_type: AutoTokenizer
|
78 |
+
train_on_inputs: false
|
79 |
+
trust_remote_code: true
|
80 |
+
val_set_size: 0.05
|
81 |
+
wandb_entity: null
|
82 |
+
wandb_mode: online
|
83 |
+
wandb_name: d39c2c44-497a-4263-b4dd-0b4ec14c1b5c
|
84 |
+
wandb_project: Gradients-On-Demand
|
85 |
+
wandb_run: your_name
|
86 |
+
wandb_runid: d39c2c44-497a-4263-b4dd-0b4ec14c1b5c
|
87 |
+
warmup_ratio: 0.05
|
88 |
+
weight_decay: 0.01
|
89 |
+
xformers_attention: true
|
90 |
+
|
91 |
+
```
|
92 |
+
|
93 |
+
</details><br>
|
94 |
+
|
95 |
+
# d39c2c44-497a-4263-b4dd-0b4ec14c1b5c
|
96 |
+
|
97 |
+
This model is a fine-tuned version of [unsloth/Qwen2.5-3B-Instruct](https://huggingface.co/unsloth/Qwen2.5-3B-Instruct) on the None dataset.
|
98 |
+
It achieves the following results on the evaluation set:
|
99 |
+
- Loss: 0.7296
|
100 |
+
|
101 |
+
## Model description
|
102 |
+
|
103 |
+
More information needed
|
104 |
+
|
105 |
+
## Intended uses & limitations
|
106 |
+
|
107 |
+
More information needed
|
108 |
+
|
109 |
+
## Training and evaluation data
|
110 |
+
|
111 |
+
More information needed
|
112 |
+
|
113 |
+
## Training procedure
|
114 |
+
|
115 |
+
### Training hyperparameters
|
116 |
+
|
117 |
+
The following hyperparameters were used during training:
|
118 |
+
- learning_rate: 0.0001
|
119 |
+
- train_batch_size: 2
|
120 |
+
- eval_batch_size: 2
|
121 |
+
- seed: 42
|
122 |
+
- distributed_type: multi-GPU
|
123 |
+
- num_devices: 4
|
124 |
+
- gradient_accumulation_steps: 16
|
125 |
+
- total_train_batch_size: 128
|
126 |
+
- total_eval_batch_size: 8
|
127 |
+
- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
|
128 |
+
- lr_scheduler_type: cosine
|
129 |
+
- lr_scheduler_warmup_steps: 2
|
130 |
+
- training_steps: 50
|
131 |
+
|
132 |
+
### Training results
|
133 |
+
|
134 |
+
| Training Loss | Epoch | Step | Validation Loss |
|
135 |
+
|:-------------:|:------:|:----:|:---------------:|
|
136 |
+
| 2.2561 | 0.0003 | 1 | 4.7806 |
|
137 |
+
| 0.6943 | 0.0066 | 25 | 0.8254 |
|
138 |
+
| 0.6894 | 0.0132 | 50 | 0.7296 |
|
139 |
+
|
140 |
+
|
141 |
+
### Framework versions
|
142 |
+
|
143 |
+
- PEFT 0.13.2
|
144 |
+
- Transformers 4.46.3
|
145 |
+
- Pytorch 2.3.1+cu121
|
146 |
+
- Datasets 3.1.0
|
147 |
+
- Tokenizers 0.20.3
|
adapter_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:949a92aac9f6bb6a0433e120d719833ee7a7345b4c70bb4a503f74b6b16c5445
|
3 |
+
size 239650666
|