End of training

README.md

@@ -19,7 +19,7 @@ axolotl version: `0.4.1`
 ```yaml
 adapter: lora
 base_model: peft-internal-testing/tiny-dummy-qwen2
-bf16: auto
+bf16: true
 chat_template: llama3
 dataset_prepared_path: null
 datasets:
@@ -36,34 +36,34 @@ datasets:
 debug: null
 deepspeed: null
 early_stopping_patience: null
-eval_max_new_tokens: 128
+eval_max_new_tokens: 64
 eval_table_size: null
-evals_per_epoch: 1
+evals_per_epoch: 0
 flash_attention: false
-fp16: null
+fp16: false
 fsdp: null
 fsdp_config: null
 gradient_accumulation_steps: 1
 gradient_checkpointing: false
-group_by_length: false
+group_by_length: true
 hub_model_id: willtensora/123e4567-e89b-12d3-a456-426614174000
 hub_repo: null
 hub_strategy: checkpoint
 hub_token: null
-learning_rate: 0.0002
+learning_rate: 0.001
 load_in_4bit: false
 load_in_8bit: false
 local_rank: null
 logging_steps: 1
-lora_alpha: 16
-lora_dropout: 0.05
+lora_alpha: 8
+lora_dropout: 0.1
 lora_fan_in_fan_out: null
 lora_model_dir: null
-lora_r: 8
+lora_r: 4
 lora_target_linear: true
-lr_scheduler: cosine
+lr_scheduler: linear
 max_steps: 1
-micro_batch_size: 1
+micro_batch_size: 4
 mlflow_experiment_name: argilla/databricks-dolly-15k-curated-en
 model_type: AutoModelForCausalLM
 num_epochs: 1
@@ -73,21 +73,21 @@ pad_to_sequence_len: true
 resume_from_checkpoint: null
 s2_attention: null
 sample_packing: false
-saves_per_epoch: 1
-sequence_len: 128
+saves_per_epoch: 0
+sequence_len: 64
 strict: false
-tf32: false
+tf32: true
 tokenizer_type: AutoTokenizer
 train_on_inputs: false
 trust_remote_code: true
-val_set_size: 0.01
+val_set_size: 0.001
 wandb_entity: null
-wandb_mode: online
+wandb_mode: disabled
 wandb_name: 123e4567-e89b-12d3-a456-426614174000
 wandb_project: Gradients-On-Demand
 wandb_run: your_name
 wandb_runid: 123e4567-e89b-12d3-a456-426614174000
-warmup_steps: 1
+warmup_steps: 0
 weight_decay: 0.0
 xformers_attention: null
 
@@ -99,7 +99,7 @@ xformers_attention: null
 
 This model is a fine-tuned version of [peft-internal-testing/tiny-dummy-qwen2](https://huggingface.co/peft-internal-testing/tiny-dummy-qwen2) on the None dataset.
 It achieves the following results on the evaluation set:
-- Loss: 11.9321
+- Loss: 11.9339
 
 ## Model description
 
@@ -118,20 +118,19 @@ More information needed
 ### Training hyperparameters
 
 The following hyperparameters were used during training:
-- learning_rate: 0.0002
-- train_batch_size: 1
-- eval_batch_size: 1
+- learning_rate: 0.001
+- train_batch_size: 4
+- eval_batch_size: 4
 - 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: 2
+- lr_scheduler_type: linear
 - training_steps: 1
 
 ### Training results
 
 | Training Loss | Epoch  | Step | Validation Loss |
 |:-------------:|:------:|:----:|:---------------:|
-| 11.9692       | 0.0001 | 1    | 11.9321         |
+| 11.9308       | 0.0003 | 1    | 11.9339         |
 
 
 ### Framework versions

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:af61f0eb3dcc19c48f6f6f43258b2369b06a62b10c0403f28a3430c0046b636c
-size 21378
+oid sha256:72cab532378323b0cc53540e5a55a19e97e56b2c647c403587d3339078717c5e
+size 15746