Model save
Browse files- README.md +67 -0
- adapter_model.safetensors +1 -1
README.md
ADDED
|
@@ -0,0 +1,67 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: bkai-foundation-models/vietnamese-llama2-7b-120GB
|
| 3 |
+
tags:
|
| 4 |
+
- generated_from_trainer
|
| 5 |
+
model-index:
|
| 6 |
+
- name: peft-bkai-alpha-32-checkpoint
|
| 7 |
+
results: []
|
| 8 |
+
library_name: peft
|
| 9 |
+
---
|
| 10 |
+
|
| 11 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 12 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 13 |
+
|
| 14 |
+
# peft-bkai-alpha-32-checkpoint
|
| 15 |
+
|
| 16 |
+
This model is a fine-tuned version of [bkai-foundation-models/vietnamese-llama2-7b-120GB](https://huggingface.co/bkai-foundation-models/vietnamese-llama2-7b-120GB) on an unknown dataset.
|
| 17 |
+
|
| 18 |
+
## Model description
|
| 19 |
+
|
| 20 |
+
More information needed
|
| 21 |
+
|
| 22 |
+
## Intended uses & limitations
|
| 23 |
+
|
| 24 |
+
More information needed
|
| 25 |
+
|
| 26 |
+
## Training and evaluation data
|
| 27 |
+
|
| 28 |
+
More information needed
|
| 29 |
+
|
| 30 |
+
## Training procedure
|
| 31 |
+
|
| 32 |
+
|
| 33 |
+
The following `bitsandbytes` quantization config was used during training:
|
| 34 |
+
- quant_method: bitsandbytes
|
| 35 |
+
- _load_in_8bit: False
|
| 36 |
+
- _load_in_4bit: True
|
| 37 |
+
- llm_int8_threshold: 6.0
|
| 38 |
+
- llm_int8_skip_modules: None
|
| 39 |
+
- llm_int8_enable_fp32_cpu_offload: False
|
| 40 |
+
- llm_int8_has_fp16_weight: False
|
| 41 |
+
- bnb_4bit_quant_type: nf4
|
| 42 |
+
- bnb_4bit_use_double_quant: True
|
| 43 |
+
- bnb_4bit_compute_dtype: float16
|
| 44 |
+
- load_in_4bit: True
|
| 45 |
+
- load_in_8bit: False
|
| 46 |
+
### Training hyperparameters
|
| 47 |
+
|
| 48 |
+
The following hyperparameters were used during training:
|
| 49 |
+
- learning_rate: 0.0002
|
| 50 |
+
- train_batch_size: 2
|
| 51 |
+
- eval_batch_size: 2
|
| 52 |
+
- seed: 42
|
| 53 |
+
- gradient_accumulation_steps: 2
|
| 54 |
+
- total_train_batch_size: 4
|
| 55 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 56 |
+
- lr_scheduler_type: cosine
|
| 57 |
+
- lr_scheduler_warmup_steps: 2
|
| 58 |
+
- num_epochs: 3
|
| 59 |
+
- mixed_precision_training: Native AMP
|
| 60 |
+
|
| 61 |
+
### Framework versions
|
| 62 |
+
|
| 63 |
+
- PEFT 0.5.0
|
| 64 |
+
- Transformers 4.38.2
|
| 65 |
+
- Pytorch 2.1.2
|
| 66 |
+
- Datasets 2.16.0
|
| 67 |
+
- Tokenizers 0.15.2
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 159967880
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:eb6ea11efffd7e3e2c1b82d0004f893f769a4c411189f8e91e8b6b5c18c94335
|
| 3 |
size 159967880
|