---
library_name: peft
license: apache-2.0
base_model: Qwen/Qwen3-1.7B
tags:
- axolotl
- base_model:adapter:Qwen/Qwen3-1.7B
- lora
- transformers
datasets:
- TeamPV/sharegpt-mistral-onr
pipeline_tag: text-generation
model-index:
- name: mistral-nemo-onr-dora-1p7
  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/axolotl-ai-cloud/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/axolotl-ai-cloud/axolotl)
<details><summary>See axolotl config</summary>

axolotl version: `0.13.0.dev0`
```yaml
base_model: Qwen/Qwen3-1.7B

# Quantization

bnb_config_kwargs:
  bnb_4bit_compute_dtype: bfloat16
  bnb_4bit_quant_type: nf4
  bnb_4bit_use_double_quant: true

datasets:
  - path: TeamPV/sharegpt-mistral-onr
    split: train
    type: chat_template
    conversation: messages  # Your dataset has 'messages' field
ds_type: json

# Use model's built-in chat template

val_set_size: 0.0
test_datasets:
  - path: TeamPV/sharegpt-mistral-onr
    split: validation
    type: chat_template
    conversation: messages
    
eval_sample_packing: false
eval_batch_size: 6
eval_steps: 30000
early_stopping_patience: 3


# Tokenization  
chat_template: tokenizer_default
sequence_len: 1200
pad_to_sequence_len: true
sample_packing: false

special_tokens:
  pad_token: "</s>"
  
# LoRA/DoRA
adapter: lora
lora_r: 32
lora_alpha: 64
lora_dropout: 0.05
lora_target_modules:
  - q_proj
  - k_proj
  - v_proj
  - o_proj
  - up_proj
  - down_proj
  - gate_proj
peft_use_dora: false
output_dir: /output/qwen1p7
use_tensorboard: true

# Training
micro_batch_size: 5
gradient_accumulation_steps: 1
num_epochs: 4
learning_rate: 0.00005
lr_scheduler: cosine
warmup_ratio: 0.10

# Optimizer
# optimizer: adamw_torch_fused
optimizer: adamw_bnb_8bit
bf16: true
fp16: false
# tf32: true

# Attention
flash_attention: true

# Memory
gradient_checkpointing: true
gradient_checkpointing_kwargs:
  use_reentrant: false
  
# Checkpointing
save_steps: 30000


save_total_limit: 2
load_best_model_at_end: true

# Logging
logging_steps: 50

# HuggingFace Hub upload
hub_model_id: TeamPV/mistral-nemo-onr-dora-1p7  # Your HF repo name
hub_strategy: end  # Options: end, every_save, checkpoint, all_checkpoints
hf_use_auth_token: true

# Optional: make repo private
```

</details><br>

# mistral-nemo-onr-dora-1p7

This model is a fine-tuned version of [Qwen/Qwen3-1.7B](https://huggingface.co/Qwen/Qwen3-1.7B) on the TeamPV/sharegpt-mistral-onr dataset.
It achieves the following results on the evaluation set:
- Loss: 1.0987
- Memory/max Active (gib): 14.15
- Memory/max Allocated (gib): 14.15
- Memory/device Reserved (gib): 14.87

## 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: 5e-05
- train_batch_size: 5
- eval_batch_size: 6
- seed: 42
- optimizer: Use OptimizerNames.ADAMW_BNB 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: 7353
- training_steps: 73530

### Training results

| Training Loss | Epoch  | Step  | Validation Loss | Active (gib) | Allocated (gib) | Reserved (gib) |
|:-------------:|:------:|:-----:|:---------------:|:------------:|:---------------:|:--------------:|
| No log        | 0      | 0     | 3.6536          | 14.08        | 14.08           | 14.15          |
| 1.0537        | 1.6319 | 30000 | 1.1174          | 14.15        | 14.15           | 14.85          |
| 0.9286        | 3.2639 | 60000 | 1.0987          | 14.15        | 14.15           | 14.87          |


### Framework versions

- PEFT 0.17.1
- Transformers 4.57.0
- Pytorch 2.7.1+cu126
- Datasets 4.0.0
- Tokenizers 0.22.1