evilfreelancer / r1_yandexgpt5-lite_lora

huggingface.co
Total runs: 14
24-hour runs: 0
7-day runs: -1
30-day runs: 9
Model's Last Updated: March 02 2025
text-generation

Introduction of r1_yandexgpt5-lite_lora

Model Details of r1_yandexgpt5-lite_lora

Russian r1 / YandexGPT-5-Lite-8B-pretrain

LoRA-адаптер для модели YandexGPT-5-Lite-8B-pretrain обученный на миксе из датасетов реализующих r1 (ризонинг) подход.

Обученная модель способна имитировать логические размышлению на русском языке по аналогии с тем, как это делает r1 от DeepSeek или o1 от OpenAI .

W&B отчёт: https://api.wandb.ai/links/evilfreelancer/zj6s02v4

Обучение производилось при помощи утилиты impruver используя конфигурацию YandexGPT/8B_lora_r1 .

На всё про всё ушло примерно 18 часов на RTX 4090, при этом понадобилось 23.5Гб видеопамяти.

Эффективный контекст 1400 токенов, так как больше не удалось уместить в 24Гб VRAM.

output_dir: ./models/YandexGPT-5-Lite_7B_lora_thinking
train_path: ./train.YandexGPT-5-Lite_7B_lora_thinking.jsonl
val_path: ./val.YandexGPT-5-Lite_7B_lora_thinking.jsonl

datasets:
  - name: mizinovmv/ru_example_DeepSeek-R1-Distill-Qwen-32B
    converter: impruver.instruction_to_messages
    add_global_bos: false
    add_global_eos: false
    mapping:
      instruction: ru_query
      output: response
  - name: lightblue/reasoning-multilingual-R1-Llama-70B-train
    converter: impruver.instruction_to_messages
    add_global_bos: false
    add_global_eos: false
    mapping:
      instruction: translated_prompt
      output: response
  - name: Pinkstack/thinking-multilingual-30-23-full-690
    converter: impruver.instruction_to_messages
    add_global_bos: false
    add_global_eos: false
  - name: Vikhrmodels/reasoning-0.01-ru
    converter: impruver.reasoning_to_messages
    add_global_bos: false
    add_global_eos: false
  - name: Vikhrmodels/russian_math
    converter: impruver.reasoning_to_messages
    add_global_bos: false
    add_global_eos: false
    mapping:
      instruction: task
      reasoning: solution
      output: short answer
  - name: kristaller486/Nebo-T1-Russian
    converter: impruver.reasoning_to_messages
    add_global_bos: false
    add_global_eos: false
    mapping:
      instruction: prompt
      reasoning: think
      output: answer

model:
  class: transformers.AutoModelForCausalLM
  name: yandex/YandexGPT-5-Lite-8B-pretrain
  load_in_4bit: true
  load_in_8bit: false
  dtype: bf16

lora:
  r: 8            # higher increases accuracy and memory
  lora_alpha: 16  # usually alpha=2*rank
  lora_dropout: 0
  bias: none
  target_modules: [ 'q_proj', 'v_proj', 'output_proj' ]
  task_type: CAUSAL_LM

tokenizer:
  class: transformers.AutoTokenizer
  name: yandex/YandexGPT-5-Lite-8B-pretrain
  max_tokens_count: 1400
  special_tokens:
    pad_token_id: 1
    pad_token: <s>

trainer:
  eval_strategy: steps
  save_strategy: steps
  eval_steps: 1000
  save_steps: 1000
  per_device_train_batch_size: 1
  per_device_eval_batch_size: 1
  gradient_accumulation_steps: 8
  logging_steps: 10
  learning_rate: 0.000005
  num_train_epochs: 2
  lr_scheduler_type: cosine
  warmup_steps: 100
  optim: adamw_torch_4bit
  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: 0.01
  torch_compile: false

Runs of evilfreelancer r1_yandexgpt5-lite_lora on huggingface.co

14
Total runs
0
24-hour runs
0
3-day runs
-1
7-day runs
9
30-day runs

More Information About r1_yandexgpt5-lite_lora huggingface.co Model

More r1_yandexgpt5-lite_lora license Visit here:

https://choosealicense.com/licenses/mit

r1_yandexgpt5-lite_lora huggingface.co

r1_yandexgpt5-lite_lora huggingface.co is an AI model on huggingface.co that provides r1_yandexgpt5-lite_lora's model effect (), which can be used instantly with this evilfreelancer r1_yandexgpt5-lite_lora model. huggingface.co supports a free trial of the r1_yandexgpt5-lite_lora model, and also provides paid use of the r1_yandexgpt5-lite_lora. Support call r1_yandexgpt5-lite_lora model through api, including Node.js, Python, http.

r1_yandexgpt5-lite_lora huggingface.co Url

https://huggingface.co/evilfreelancer/r1_yandexgpt5-lite_lora

evilfreelancer r1_yandexgpt5-lite_lora online free

r1_yandexgpt5-lite_lora huggingface.co is an online trial and call api platform, which integrates r1_yandexgpt5-lite_lora's modeling effects, including api services, and provides a free online trial of r1_yandexgpt5-lite_lora, you can try r1_yandexgpt5-lite_lora online for free by clicking the link below.

evilfreelancer r1_yandexgpt5-lite_lora online free url in huggingface.co:

https://huggingface.co/evilfreelancer/r1_yandexgpt5-lite_lora

r1_yandexgpt5-lite_lora install

r1_yandexgpt5-lite_lora is an open source model from GitHub that offers a free installation service, and any user can find r1_yandexgpt5-lite_lora on GitHub to install. At the same time, huggingface.co provides the effect of r1_yandexgpt5-lite_lora install, users can directly use r1_yandexgpt5-lite_lora installed effect in huggingface.co for debugging and trial. It also supports api for free installation.

r1_yandexgpt5-lite_lora install url in huggingface.co:

https://huggingface.co/evilfreelancer/r1_yandexgpt5-lite_lora

Url of r1_yandexgpt5-lite_lora

r1_yandexgpt5-lite_lora huggingface.co Url

Provider of r1_yandexgpt5-lite_lora huggingface.co

evilfreelancer
ORGANIZATIONS

Other API from evilfreelancer