File size: 1,729 Bytes
a0f43c2 7a7de9f a0f43c2 45871cc a0f43c2 ddcbd3c e7805df 7a7de9f c7692fe 7a7de9f c7692fe 7a7de9f |
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 |
---
base_model: ai-forever/ruGPT-3.5-13B
library_name: peft
license: mit
datasets:
- korotkov/glaive-function-calling-v2-ru-parsed
language:
- ru
tags:
- impruver
- russian
- function call
- lora
pipeline_tag: text-generation
---
# ruGPT-3.5-13B / function call
LoRA адаптер для ruGPT3.5-13B обученный на датасете function call.
Конфигурация: https://github.com/EvilFreelancer/impruver/blob/main/recipes/configs/ruGPT-3.5/13B_lora_fc.yaml
Адаптер обучался на 1x RTX 4090, для этого потребовалось примерно 20Gb VRAM и заняло 11h 8m.
```yaml
output_dir: ./models/ruGPT35_13B_lora_fc
train_path: ./train.ruGPT35_13B_fc.jsonl
val_path: ./val.ruGPT35_13B_fc.jsonl
datasets:
- name: korotkov/glaive-function-calling-v2-ru-parsed
split: train
model:
class: transformers.AutoModelForCausalLM
name: ai-forever/ruGPT-3.5-13B
load_in_4bit: true
load_in_8bit: false
dtype: bf16
lora:
r: 16
lora_alpha: 16
lora_dropout: 0.05
bias: none
target_modules: [ c_attn ]
task_type: CAUSAL_LM
tokenizer:
class: transformers.AutoTokenizer
name: ai-forever/ruGPT-3.5-13B
max_tokens_count: 1200
trainer:
eval_strategy: steps
save_strategy: steps
eval_steps: 100
save_steps: 100
per_device_train_batch_size: 1
per_device_eval_batch_size: 1
gradient_accumulation_steps: 128
logging_steps: 1
learning_rate: 0.0002
num_train_epochs: 2
lr_scheduler_type: cosine
warmup_steps: 16
optim: adamw_8bit
metric_for_best_model: eval_loss
load_best_model_at_end: true
save_total_limit: 2
seed: 42
remove_unused_columns: false
max_grad_norm: 1.0
weight_decay: 0.08
torch_compile: false
``` |