metadata
library_name: peft
license: mit
base_model: ArliAI/GLM-4.5-Air-Derestricted
tags:
- axolotl
- base_model:adapter:ArliAI/GLM-4.5-Air-Derestricted
- lora
- transformers
datasets:
- data/data.jsonl
pipeline_tag: text-generation
model-index:
- name: output
results: []
See axolotl config
axolotl version: 0.13.0.dev0
# Weights and Biases logging config
wandb_project: Test Run _Q3
wandb_name: "0.1"
# Model architecture config
base_model: ArliAI/GLM-4.5-Air-Derestricted
model_type: AutoModelForCausalLM
# Model checkpointing config
output_dir: ./output
saves_per_epoch: 1
save_safetensors: true
save_total_limit: 1
# Mixed precision training config
bf16: true
fp16: false
tf32: false
# Model loading config
load_in_8bit: true
load_in_4bit: false
strict: false
# Sequence config
sequence_len: 8192
s2_attention: false
sample_packing: true
eval_sample_packing: true
pad_to_sequence_len: true
train_on_inputs: false
group_by_length: false
# QLoRA adapter config
adapter: lora
lora_r: 64
lora_alpha: 32
lora_dropout: 0.05
peft_use_dora: false
lora_target_modules:
- gate_proj
- down_proj
- up_proj
- q_proj
- v_proj
- k_proj
- o_proj
# Dataset config
datasets:
- path: data/data.jsonl
type: chat_template
field_messages: conversations
message_field_role: role
message_field_content: content
# Training hyperparameters
num_epochs: 2
gradient_accumulation_steps: 2
micro_batch_size: 2
eval_batch_size: 1
warmup_steps: 500
optimizer: paged_adamw_8bit
lr_scheduler: cosine
learning_rate: 1e-5
loraplus_lr_ratio: 8
cosine_min_lr_ratio: 0.1
weight_decay: 0.1
max_grad_norm: 1
logging_steps: 10
# === Plugins ===
plugins:
- axolotl.integrations.liger.LigerPlugin
- axolotl.integrations.cut_cross_entropy.CutCrossEntropyPlugin
liger_rope: true
liger_rms_norm: true
liger_glu_activation: true
cut_cross_entropy: true
# Model optimization
gradient_checkpointing: offload
xformers_attention: false
flash_attention: true
sdp_attention: false
# Loss monitoring config
early_stopping_patience: false
loss_watchdog_threshold: 100.0
loss_watchdog_patience: 3
# Debug config
debug: false
seed: 42
deepspeed: deepspeed_configs/zero2.json
output
This model is a fine-tuned version of ArliAI/GLM-4.5-Air-Derestricted on the data/data.jsonl dataset.
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: 1e-05
- train_batch_size: 2
- eval_batch_size: 1
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- gradient_accumulation_steps: 2
- total_train_batch_size: 16
- total_eval_batch_size: 4
- optimizer: Use OptimizerNames.PAGED_ADAMW_8BIT 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: 500
- training_steps: 48
Training results
Framework versions
- PEFT 0.18.0
- Transformers 4.57.1
- Pytorch 2.8.0+cu128
- Datasets 4.4.1
- Tokenizers 0.22.1