Views
No views yet
0.4.01base_model: aurora-m/aurora-m-v0.1 # this can be swapped for mdel model when the model is released
2model_type: AutoModelForCausalLM
3tokenizer_type: AutoTokenizer
4is_llama_derived_model: false
5
6load_in_8bit: false # when this is true inference quality is terrible
7load_in_4bit: false
8strict: false
9
10datasets:
11 - path: /workspace/axolotl-mdel/mtg.txt # change this to where your dataset is
12 type: completion # change this to 'alpaca' if you are using alpaca formatting
13
14lora_modules_to_save:
15 - embed_tokens
16 - lm_head
17
18dataset_prepared_path:
19val_set_size: 0.05
20output_dir: ./lora-out
21
22sequence_len: 4096 # this can be tweaked for efficiency
23sample_packing: true
24pad_to_sequence_len: true
25
26adapter: lora
27lora_model_dir:
28lora_r: 32
29lora_alpha: 16
30lora_dropout: 0.05
31lora_target_linear: true
32lora_fan_in_fan_out:
33
34wandb_project: mtg-aurora-experiement # give this a name
35wandb_entity:
36wandb_watch:
37wandb_name:
38wandb_log_model:
39
40gradient_accumulation_steps: 2 # this can be tweaked for efficiency
41micro_batch_size: 1 # this can be tweaked for efficiency
42num_epochs: 1 # this can be experimented with
43optimizer: adamw_bnb_8bit
44lr_scheduler: cosine
45learning_rate: 0.0002
46
47train_on_inputs: true
48group_by_length: false
49bf16: true
50fp16: false
51tf32: false
52
53gradient_checkpointing: true
54early_stopping_patience:
55resume_from_checkpoint:
56local_rank:
57logging_steps: 1
58xformers_attention:
59flash_attention: false # when this is true, inference quality is terrible
60s2_attention:
61
62warmup_steps: 10 # this can be tweaked for efficiency
63evals_per_epoch: 10 # this can be tweaked for efficiency
64eval_table_size:
65eval_table_max_new_tokens: 128
66saves_per_epoch: 1
67debug:
68deepspeed:
69weight_decay: 0.0
70fsdp:
71fsdp_config:
72special_tokens:
73 pad_token: "<|endoftext|>"
74 eos_token: "<|endoftext|>"
75| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 4.2833 | 0.0 | 1 | 4.0839 |
| 2.1947 | 0.1 | 25 | 1.9886 |
| 1.2659 | 0.21 | 50 | 1.1937 |
| 1.0662 | 0.31 | 75 | 1.0060 |
| 0.9538 | 0.41 | 100 | 0.9172 |
| 0.9232 | 0.52 | 125 | 0.8603 |
| 0.8546 | 0.62 | 150 | 0.8237 |
| 0.8223 | 0.73 | 175 | 0.8049 |
| 0.8546 | 0.83 | 200 | 0.7979 |
| 0.8995 | 0.93 | 225 | 0.7945 |