Hugging Face's logo

Crystalcareai
/
CrystalMiniCPM

Text Generation
Transformers
PyTorch
TensorBoard
Safetensors
minicpm
Generated from Trainer
conversational
custom_code
Model card Files Metrics Community
CrystalMiniCPM
File size: 2,978 Bytes 
f712e6f
206b412
 
 
 
 
 
f712e6f
206b412
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
 
---
base_model: openbmb/MiniCPM-2B-sft-bf16
tags:
- generated_from_trainer
model-index:
- name: qlora-out
  results: []
---

<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->

[<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)
<details><summary>See axolotl config</summary>

axolotl version: `0.4.0`
```yaml
base_model: openbmb/MiniCPM-2B-sft-bf16
load_in_8bit: false
load_in_4bit: false
strict: false
push_dataset_to_hub:
datasets:
  - path: teknium/GPT4-LLM-Cleaned
    type: alpaca
dataset_prepared_path:
val_set_size: 0.05
adapter:
lora_model_dir:
sequence_len: 4096
max_packed_sequence_len:
lora_r: 8
lora_alpha: 32
lora_dropout: 0.05
lora_target_modules:
lora_target_linear: true
lora_fan_in_fan_out:
wandb_project:
wandb_entity:
wandb_watch:
wandb_name:
wandb_log_model:
output_dir: ./qlora-out
gradient_accumulation_steps: 2
micro_batch_size: 2
num_epochs: 1.5
optimizer: paged_adamw_8bit
torchdistx_path:
lr_scheduler: cosine
learning_rate: 0.0001
train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: true
gradient_checkpointing: true
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention: true
flash_attention:
gptq_groupsize:
gptq_model_v1:
warmup_steps: 10
evals_per_epoch: 2
saves_per_epoch: 1
debug:
deepspeed:
weight_decay: 0.1
fsdp:
fsdp_config:
special_tokens:
trust_remote_code: true

```

</details><br>

# qlora-out

This model is a fine-tuned version of [openbmb/MiniCPM-2B-sft-bf16](https://huggingface.co/openbmb/MiniCPM-2B-sft-bf16) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 1.0525

## Model description

More information needed

## Intended uses & limitations

More information needed

## Training and evaluation data

More information needed

## Training procedure

### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 0.0001
- train_batch_size: 2
- eval_batch_size: 2
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- gradient_accumulation_steps: 2
- total_train_batch_size: 16
- total_eval_batch_size: 8
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 10
- num_epochs: 1.5

### Training results

| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:-----:|:----:|:---------------:|
| 1.0903        | 0.0   | 1    | 1.7199          |
| 0.8959        | 0.5   | 1620 | 1.1007          |
| 0.995         | 1.0   | 3240 | 1.0342          |
| 0.864         | 1.5   | 4860 | 1.0525          |


### Framework versions

- Transformers 4.38.0.dev0
- Pytorch 2.1.2+cu118
- Datasets 2.16.1
- Tokenizers 0.15.0