QuantFactory Banner

QuantFactory/sarashina2.1-1b-sft-GGUF

This is quantized version of Aratako/sarashina2.1-1b-sft created using llama.cpp

Original Model Card

sarashina2.1-1b-sft

sbintuitions/sarashina2.1-1bใ‚’ๆ‰‹ๅ…ƒใฎใƒ‡ใƒผใ‚ฟใงInstruction Tuningใ—ใŸใƒขใƒ‡ใƒซใงใ™ใ€‚

Sarashina is licensed under the Sarashina Model NonCommercial License Agreement, Copyright ยฉSB Intuitions Corp. All Rights Reserved.

This model is built with Qwen.

ไฝฟใ„ๆ–น

from transformers import AutoModelForCausalLM, AutoTokenizer, TextStreamer

model = AutoModelForCausalLM.from_pretrained("Aratako/sarashina2.1-1b-sft", device_map="auto", torch_dtype="auto")
tokenizer = AutoTokenizer.from_pretrained("Aratako/sarashina2.1-1b-sft")
streamer = TextStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True)

messages = [
  {"role": "user", "content": "ใ‚ฝใƒ•ใƒˆใƒใƒณใ‚ฏๆ ชๅผไผš็คพใซใคใ„ใฆๆ•™ใˆใฆใใ ใ•ใ„ใ€‚"}
]


input_ids = tokenizer.apply_chat_template(messages, add_generation_prompt=True, return_tensors="pt").to(model.device)
output_ids = model.generate(input_ids,
                            max_new_tokens=1024,
                            do_sample=False,
                            streamer=streamer,
                            eos_token_id=tokenizer.eos_token_id,
                           )
ใ‚ฝใƒ•ใƒˆใƒใƒณใ‚ฏๆ ชๅผไผš็คพใฏใ€ๆ—ฅๆœฌใฎๅคงๆ‰‹้€šไฟกไผš็คพใฎไธ€ใคใงใ€ไธปใซๆบๅธฏ้›ป่ฉฑใ‚„ใ‚คใƒณใ‚ฟใƒผใƒใƒƒใƒˆใ‚ตใƒผใƒ“ใ‚นใ€ใใ—ใฆๆง˜ใ€…ใชใƒ‡ใ‚ธใ‚ฟใƒซ่ฃฝๅ“ใ‚„ใ‚ตใƒผใƒ“ใ‚นใ‚’ๆไพ›ใ—ใฆใ„ใพใ™ใ€‚

1. **ไบ‹ๆฅญๅ†…ๅฎน**:
   - ๆบๅธฏ้›ป่ฉฑไบ‹ๆฅญ: ใ‚ฝใƒ•ใƒˆใƒใƒณใ‚ฏใƒขใƒใ‚คใƒซ๏ผˆ็พใƒปใ‚ฝใƒ•ใƒˆใƒใƒณใ‚ฏ๏ผ‰ใจใ—ใฆใ€ๆบๅธฏ้›ป่ฉฑใ‚„ใ‚นใƒžใƒผใƒˆใƒ•ใ‚ฉใƒณใฎ่ฒฉๅฃฒใ‚’่กŒใฃใฆใ„ใพใ™ใ€‚
   - ใ‚คใƒณใ‚ฟใƒผใƒใƒƒใƒˆใ‚ตใƒผใƒ“ใ‚น: ใ‚ฝใƒ•ใƒˆใƒใƒณใ‚ฏBB๏ผˆ็พใƒปใ‚ฝใƒ•ใƒˆใƒใƒณใ‚ฏBB๏ผ‰ใจใ—ใฆใ€ใ‚คใƒณใ‚ฟใƒผใƒใƒƒใƒˆๆŽฅ็ถšใ‚ตใƒผใƒ“ใ‚นใ‚„ISPไบ‹ๆฅญใ‚’่กŒใฃใฆใ„ใพใ™ใ€‚
   - ใƒ‡ใ‚ธใ‚ฟใƒซ่ฃฝๅ“: ใ‚นใƒžใƒผใƒˆใƒ•ใ‚ฉใƒณใ‚„ใ‚ฟใƒ–ใƒฌใƒƒใƒˆใ€ใ‚ฆใ‚งใ‚ขใƒฉใƒ–ใƒซใƒ‡ใƒใ‚คใ‚นใชใฉใฎ้–‹็™บใƒป่ฒฉๅฃฒใ‚‚่กŒใฃใฆใ„ใพใ™ใ€‚
   - ใ‚ฏใƒฉใ‚ฆใƒ‰ใ‚ตใƒผใƒ“ใ‚น: ใ‚ฝใƒ•ใƒˆใƒใƒณใ‚ฏใ‚ฏใƒฉใ‚ฆใƒ‰ใจใ—ใฆใ€ใ‚ฏใƒฉใ‚ฆใƒ‰ใ‚ตใƒผใƒ“ใ‚นใ‚’ๆไพ›ใ—ใฆใ„ใพใ™ใ€‚
   - ้‡‘่žใ‚ตใƒผใƒ“ใ‚น: ใ‚ฝใƒ•ใƒˆใƒใƒณใ‚ฏใƒ•ใ‚กใ‚คใƒŠใƒณใ‚นใจใ—ใฆใ€ใ‚ฏใƒฌใ‚ธใƒƒใƒˆใ‚ซใƒผใƒ‰ใ‚„ใƒญใƒผใƒณใชใฉใฎ้‡‘่žใ‚ตใƒผใƒ“ใ‚นใ‚’ๆไพ›ใ—ใฆใ„ใพใ™ใ€‚

2. **ใ‚ฐใƒซใƒผใƒ—ไผš็คพ**:
   - ใ‚ฝใƒ•ใƒˆใƒใƒณใ‚ฏใ‚ฐใƒซใƒผใƒ—: ใ‚ฝใƒ•ใƒˆใƒใƒณใ‚ฏๆ ชๅผไผš็คพใฎ่ฆชไผš็คพใงใ‚ใ‚Šใ€้€šไฟกไบ‹ๆฅญใ ใ‘ใงใชใใ€ใƒ†ใ‚ฏใƒŽใƒญใ‚ธใƒผใ€ใƒกใƒ‡ใ‚ฃใ‚ขใ€้‡‘่žใชใฉๅคšๅฒใซใ‚ใŸใ‚‹ไบ‹ๆฅญใ‚’ๅฑ•้–‹ใ—ใฆใ„ใพใ™ใ€‚
   - ใ‚ฝใƒ•ใƒˆใƒใƒณใ‚ฏBB: ใ‚คใƒณใ‚ฟใƒผใƒใƒƒใƒˆๆŽฅ็ถšใ‚ตใƒผใƒ“ใ‚นใ‚„ISPไบ‹ๆฅญใ‚’่กŒใ†ๅญไผš็คพใงใ™ใ€‚
   - ใ‚ฝใƒ•ใƒˆใƒใƒณใ‚ฏใƒขใƒใ‚คใƒซ: ๆบๅธฏ้›ป่ฉฑไบ‹ๆฅญใ‚’่กŒใ†ๅญไผš็คพใงใ™ใ€‚
   - ใ‚ฝใƒ•ใƒˆใƒใƒณใ‚ฏใ‚ณใƒžใƒผใ‚น&ใ‚ตใƒผใƒ“ใ‚น: ใ‚ณใƒžใƒผใ‚นไบ‹ๆฅญใ‚’่กŒใ†ๅญไผš็คพใงใ™ใ€‚
   - ใ‚ฝใƒ•ใƒˆใƒใƒณใ‚ฏใƒ†ใ‚ฏใƒŽใƒญใ‚ธใƒผ: ITใ‚ฝใƒชใƒฅใƒผใ‚ทใƒงใƒณไบ‹ๆฅญใ‚’่กŒใ†ๅญไผš็คพใงใ™ใ€‚

3. **็‰นๅพด**:
   - ใ‚ฝใƒ•ใƒˆใƒใƒณใ‚ฏใฏใ€้€šไฟกไบ‹ๆฅญใ ใ‘ใงใชใใ€ใƒ†ใ‚ฏใƒŽใƒญใ‚ธใƒผใ‚„ใƒกใƒ‡ใ‚ฃใ‚ขใ€้‡‘่žใชใฉๅคšๅฒใซใ‚ใŸใ‚‹ๅˆ†้‡Žใงไบ‹ๆฅญใ‚’ๅฑ•้–‹ใ—ใฆใŠใ‚Šใ€ใใฎ่ฆๆจกใจๅฝฑ้ŸฟๅŠ›ใฏ้žๅธธใซๅคงใใ„ใงใ™ใ€‚
   - ใ€Œใ‚ฝใƒ•ใƒˆใƒใƒณใ‚ฏใ‚ฐใƒซใƒผใƒ—ใ€ใจใ—ใฆใ€้€šไฟกไบ‹ๆฅญใ ใ‘ใงใชใใ€ใƒ†ใ‚ฏใƒŽใƒญใ‚ธใƒผใ€ใƒกใƒ‡ใ‚ฃใ‚ขใ€้‡‘่žใชใฉๅคšๅฒใซใ‚ใŸใ‚‹ไบ‹ๆฅญใ‚’ๅฑ•้–‹ใ—ใฆใ„ใพใ™ใ€‚
   - ใ€ŒPepperใ€ใชใฉใฎใƒญใƒœใƒƒใƒˆ้–‹็™บใ‚„ใ€AI๏ผˆไบบๅทฅ็Ÿฅ่ƒฝ๏ผ‰ๆŠ€่ก“ใ‚’ๆดป็”จใ—ใŸใ‚ตใƒผใƒ“ใ‚น้–‹็™บใซใ‚‚ๅŠ›ใ‚’ๅ…ฅใ‚Œใฆใ„ใพใ™ใ€‚

4. **็คพไผš่ฒข็Œฎๆดปๅ‹•**:
   - ใ‚ฝใƒ•ใƒˆใƒใƒณใ‚ฏใฏใ€็คพไผš่ฒข็Œฎๆดปๅ‹•ใซใ‚‚็ฉๆฅต็š„ใซๅ–ใ‚Š็ต„ใ‚“ใงใŠใ‚Šใ€ๆ•™่‚ฒๆ”ฏๆดใ‚„็ฝๅฎณๆ™‚ใฎๆ•‘ๆดๆดปๅ‹•ใชใฉใ€ๆง˜ใ€…ใชๅฝขใง็คพไผšใซ่ฒข็Œฎใ—ใฆใ„ใพใ™ใ€‚

ใ‚ฝใƒ•ใƒˆใƒใƒณใ‚ฏใฏใ€้€šไฟกๆฅญ็•Œใ ใ‘ใงใชใใ€ๅน…ๅบƒใ„ๅˆ†้‡Žใง้ฉๆ–ฐ็š„ใชใ‚ตใƒผใƒ“ใ‚นใ‚’ๆไพ›ใ—ใ€ๆ—ฅๆœฌใฎ็คพไผšใ‚„็ตŒๆธˆใซๅคงใใชๅฝฑ้Ÿฟใ‚’ไธŽใˆใฆใ„ใพใ™ใ€‚

ๅฟœ็ญ”ไพ‹

ๅ…ฅๅŠ›๏ผˆelyza/ELYZA-tasks-100ใ‚ˆใ‚Š๏ผ‰

<|im_start|>user
ไป•ไบ‹ใฎ็†ฑๆ„ใ‚’ๅ–ใ‚Šๆˆปใ™ใŸใ‚ใฎใ‚ขใ‚คใƒ‡ใ‚ขใ‚’5ใคๆŒ™ใ’ใฆใใ ใ•ใ„ใ€‚<|im_end|>
<|im_start|>assistant

ๅ‡บๅŠ›

ใ‚‚ใกใ‚ใ‚“ใ€ไปฅไธ‹ใซใ„ใใคใ‹ใฎใ‚ขใ‚คใƒ‡ใ‚ขใ‚’ๆๆกˆใ—ใพใ™๏ผš

1. ็›ฎๆจ™่จญๅฎš: ๆ˜Ž็ขบใช็›ฎๆจ™ใ‚’่จญๅฎšใ—ใ€ใใ‚Œใ‚’้”ๆˆใ™ใ‚‹ใŸใ‚ใฎๅ…ทไฝ“็š„ใชใ‚นใƒ†ใƒƒใƒ—ใ‚’่จˆ็”ปใ—ใพใ™ใ€‚ใ“ใ‚Œใซใ‚ˆใ‚Šใ€ไป•ไบ‹ใซๅฏพใ™ใ‚‹ใƒขใƒใƒ™ใƒผใ‚ทใƒงใƒณใŒ้ซ˜ใพใ‚Šใพใ™ใ€‚

2. ไผ‘ๆ†ฉใ‚’ๅ–ใ‚‹: ้•ทๆ™‚้–“ๅƒใใจ้›†ไธญๅŠ›ใŒไฝŽไธ‹ใ—ใพใ™ใ€‚ๅฎšๆœŸ็š„ใซไผ‘ๆ†ฉใ‚’ๅ–ใ‚Šใ€ใƒชใƒ•ใƒฌใƒƒใ‚ทใƒฅใ™ใ‚‹ใ“ใจใŒ้‡่ฆใงใ™ใ€‚

3. ๆ–ฐใ—ใ„ใ‚นใ‚ญใƒซใ‚’ๅญฆใถ: ๆ–ฐใ—ใ„ๆŠ€่ก“ใ‚„็Ÿฅ่ญ˜ใ‚’็ฟ’ๅพ—ใ™ใ‚‹ใ“ใจใงใ€่‡ชๅทฑๆˆ้•ทใ‚’ๆ„Ÿใ˜ใ€ใใ‚ŒใŒไป•ไบ‹ใธใฎๆƒ…็†ฑใ‚’ๅ†ใณๅผ•ใๅ‡บใ™ใ“ใจใŒใงใใพใ™ใ€‚

4. ๅฅๅบท็š„ใชใƒฉใ‚คใƒ•ใ‚นใ‚ฟใ‚คใƒซใ‚’็ถญๆŒใ™ใ‚‹: ้ฉๅบฆใช้‹ๅ‹•ใ‚„ๅๅˆ†ใช็ก็œ ใฏใ€็ฒพ็ฅž็š„ใชๅฅๅบทใ‚’็ถญๆŒใ—ใ€ใ‚จใƒใƒซใ‚ฎใƒผใƒฌใƒ™ใƒซใ‚’้ซ˜ใ‚ใพใ™ใ€‚

5. ใƒใƒƒใƒˆใƒฏใƒผใ‚ญใƒณใ‚ฐ: ๅŒๅƒšใ‚„ไป–ใฎๅฐ‚้–€ๅฎถใจใฎไบคๆตใ‚’้€šใ˜ใฆๆ–ฐใŸใช่ฆ–็‚นใ‚’ๅพ—ใ‚‹ใ“ใจใŒใงใใพใ™ใ€‚ใ“ใ‚Œใฏใ‚ใชใŸใฎ่ฆ–้‡Žใ‚’ๅบƒใ’ใ€ๆ–ฐใŸใชๆฉŸไผšใ‚’่ฆ‹ใคใ‘ใ‚‹ๆ‰‹ๅŠฉใ‘ใซใชใ‚Šใพใ™ใ€‚

ใƒฉใ‚คใ‚ปใƒณใ‚น

ๅ…ƒใƒขใƒ‡ใƒซใ‹ใ‚‰Sarashina Model NonCommercial License Agreementใ‚’็ถ™ๆ‰ฟใ—ใพใ™ใ€‚ๅˆฉ็”จใซใ‚ใŸใฃใฆใฏๆง˜ใ€…ใชๅˆถ็ด„ใŒใ‚ใ‚‹ใฎใงๅๅˆ†ใซใ”ๆณจๆ„ใใ ใ•ใ„ใ€‚

ๆณจๆ„

  • ๆœฌใƒขใƒ‡ใƒซใฎ่ฃฝไฝœ่€…ใฏๅ…ƒใƒขใƒ‡ใƒซใฎ่ฃฝไฝœ่€…ใงใ‚ใ‚‹SB Intuitionsๆ ชๅผไผš็คพใจใฏไธ€ๅˆ‡้–ขไฟ‚ใ‚ใ‚Šใพใ›ใ‚“ใ€‚ๆœฌใƒขใƒ‡ใƒซใซไฝ•ใ‹ๅ•้กŒใŒใ‚ใฃใŸๅ ดๅˆใฏ่ฃฝไฝœ่€…ใงใ‚ใ‚‹็งๅฎ›ใซใ”้€ฃ็ตกใใ ใ•ใ„ใ€‚
  • ใƒขใƒ‡ใƒซใฎๅฎ‰ๅ…จๆ€งใฏๆคœ่จผใ•ใ‚Œใฆใ„ใพใ›ใ‚“ใ€‚ใ”ไบ†ๆ‰ฟใใ ใ•ใ„ใ€‚

Built with Axolotl

See axolotl config

axolotl version: 0.5.2

base_model: sbintuitions/sarashina2.1-1b
model_type: AutoModelForCausalLM
tokenizer_type: AutoTokenizer

hub_model_id: Aratako/sarashina2.1-1b-sft
hub_strategy: "end"
push_dataset_to_hub:
hf_use_auth_token: true

plugins:
  - axolotl.integrations.liger.LigerPlugin
liger_cross_entropy: false
liger_rope: true
liger_rms_norm: true
liger_swiglu: true
liger_fused_linear_cross_entropy: true

load_in_8bit: false
load_in_4bit: false
strict: false

chat_template: chatml

datasets:
  - path: Aratako/Magpie-Tanuki-Qwen2.5-72B-Answered
    type: chat_template
    field_messages: messages
    message_field_role: role
    message_field_content: content
  - path: Aratako/magpie-qwen2.5-32b-reasoning-100k-formatted
    type: chat_template
    field_messages: conversations
    message_field_role: role
    message_field_content: content
  - path: Aratako/Open-Platypus-Japanese-masked-formatted
    type: chat_template
    field_messages: conversations
    message_field_role: role
    message_field_content: content
  - path: kanhatakeyama/wizardlm8x22b-logical-math-coding-sft_additional-ja
    type: chat_template
    field_messages: messages
    message_field_role: role
    message_field_content: content
  - path: kanhatakeyama/ramdom-to-fixed-multiturn-Calm3
    split: 20240806filtered
    type: chat_template
    field_messages: messages
    message_field_role: role
    message_field_content: content
  - path: llm-jp/magpie-sft-v1.0
    type: chat_template
    field_messages: conversations
    message_field_role: role
    message_field_content: content
  - path: Aratako/aya-ja-evol-instruct-calm3-dpo-masked-sft
    type: chat_template
    field_messages: messages
    message_field_role: role
    message_field_content: content
  - path: Aratako/aya-ja-nemotron-dpo-masked-sft
    type: chat_template
    field_messages: messages
    message_field_role: role
    message_field_content: content
  - path: Aratako/Synthetic-JP-EN-Coding-Dataset-801k
    split: "train[0:50000]"
    type: chat_template
    field_messages: messages
    message_field_role: role
    message_field_content: content
  - path: Aratako/orca-agentinstruct-1M-v1-selected-2
    type: chat_template
    field_messages: messages
    message_field_role: role
    message_field_content: content
  - path: Aratako/Synthetic-JP-EN-Coding-Dataset-801k-50k
    type: chat_template
    field_messages: messages
    message_field_role: role
    message_field_content: content

shuffle_merged_datasets: true
dataset_prepared_path: /workspace/data/fft-data-sarashina
val_set_size: 0.002
output_dir: /workspace/data/1b-fft-out

sequence_len: 4096
sample_packing: true
eval_sample_packing: false
pad_to_sequence_len: true

adapter:
lora_model_dir:
lora_r:
lora_alpha:
lora_dropout:
lora_target_linear:
lora_fan_in_fan_out:

wandb_project: 1b-fft
wandb_entity: aratako-lm
wandb_watch:
wandb_name: fft-attempt-1
wandb_log_model:

gradient_accumulation_steps: 4
micro_batch_size: 8
num_epochs: 2
optimizer: adamw_torch
lr_scheduler: cosine
cosine_min_lr_ratio: 0.1
learning_rate: 0.00002

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: false
early_stopping_patience:
auto_resume_from_checkpoints: true
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

save_strategy: steps
save_steps: 100
save_total_limit: 1

warmup_steps: 20
eval_steps: 100
eval_batch_size: 1
eval_table_size:
eval_max_new_tokens:
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero1.json
weight_decay: 0.01
fsdp:
fsdp_config:
special_tokens:
  pad_token: <pad>

tokens:
  - "<|im_start|>"
  - "<|im_end|>"

sarashina2.1-1b-sft

This model is a fine-tuned version of sbintuitions/sarashina2.1-1b on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.9366

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: 2e-05
  • train_batch_size: 8
  • eval_batch_size: 1
  • seed: 42
  • distributed_type: multi-GPU
  • num_devices: 8
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 256
  • total_eval_batch_size: 8
  • optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
  • lr_scheduler_type: cosine
  • lr_scheduler_warmup_steps: 20
  • num_epochs: 2

Training results

Training Loss Epoch Step Validation Loss
1.2935 0.0015 1 1.4733
0.985 0.1515 100 1.0491
0.9131 0.3029 200 1.0156
0.9174 0.4544 300 0.9935
0.9257 0.6058 400 0.9806
0.869 0.7573 500 0.9694
0.8874 0.9087 600 0.9608
0.8041 1.0594 700 0.9557
0.8348 1.2109 800 0.9512
0.8353 1.3624 900 0.9466
0.8145 1.5138 1000 0.9432
0.8057 1.6653 1100 0.9400
0.838 1.8167 1200 0.9381
0.8446 1.9682 1300 0.9366

Framework versions

  • Transformers 4.46.3
  • Pytorch 2.3.1+cu121
  • Datasets 3.1.0
  • Tokenizers 0.20.3
Downloads last month
4
GGUF
Model size
1B params
Architecture
llama
Hardware compatibility
Log In to add your hardware

2-bit

3-bit

4-bit

5-bit

6-bit

8-bit

Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for QuantFactory/sarashina2.1-1b-sft-GGUF

Quantized
(5)
this model