File size: 5,901 Bytes
4fb7979
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
be9e137
4fb7979
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
be9e137
4fb7979
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
be9e137
4fb7979
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
be9e137
4fb7979
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
_n_gpu: 4
accelerator_config:
  dispatch_batches: null
  even_batches: true
  gradient_accumulation_kwargs: null
  non_blocking: false
  split_batches: false
  use_configured_state: false
  use_seedable_sampler: true
activation_offloading: false
adafactor: false
adam_beta1: 0.9
adam_beta2: 0.999
adam_epsilon: 1.0e-08
assistant_only_loss: false
auto_find_batch_size: false
auto_infer_class_weights: false
average_tokens_across_devices: true
batch_eval_metrics: false
bf16: true
bf16_full_eval: false
chat_template_path: null
chosen_column: null
class_weights: null
completion_only_loss: null
data_dir: /project2/jonmay_1426/ashokd/RAU/llm-utils/data
data_seed: null
dataloader_drop_last: false
dataloader_num_workers: 0
dataloader_persistent_workers: false
dataloader_pin_memory: true
dataloader_prefetch_factor: null
dataset_kwargs: null
dataset_num_proc: null
dataset_text_field: text
ddp_backend: null
ddp_broadcast_buffers: null
ddp_bucket_cap_mb: null
ddp_find_unused_parameters: null
ddp_timeout: 1800
debug: []
deepspeed: null
disable_tqdm: false
do_eval: true
do_lora: true
do_predict: false
do_train: false
early_stopping_patience: null
early_stopping_threshold: 0.0
env_dir: /project2/jonmay_1426/ashokd/llm-utils/env/.venv/
eos_token: null
eval_accumulation_steps: null
eval_delay: 0
eval_do_concat_batches: true
eval_max_new_tokens: 512
eval_on_start: false
eval_packing: null
eval_steps: 500
eval_strategy: !!python/object/apply:transformers.trainer_utils.IntervalStrategy
- steps
eval_use_gather_object: false
evaluate_before_training: false
figure_dir: results/figures
figure_force_save: true
fp16: false
fp16_backend: auto
fp16_full_eval: false
fp16_opt_level: O1
fsdp: []
fsdp_config:
  min_num_params: 0
  xla: false
  xla_fsdp_grad_ckpt: false
  xla_fsdp_v2: false
fsdp_min_num_params: 0
fsdp_transformer_layer_cls_to_wrap: null
full_determinism: false
ga_forget_column: forget
gradient_accumulation_steps: 1
gradient_checkpointing: true
gradient_checkpointing_kwargs: null
greater_is_better: null
group_by_length: false
half_precision_backend: auto
hub_always_push: false
hub_model_id: Llama-3.1-8B-Instruct_rt_forget-full
hub_private_repo: null
hub_revision: null
hub_strategy: !!python/object/apply:transformers.trainer_utils.HubStrategy
- every_save
hub_token: null
huggingface_hub_username: Anonymous
ignore_data_skip: false
image_input_column: image
include_for_metrics: []
include_inputs_for_metrics: false
include_num_input_tokens_seen: 'no'
include_tokens_per_second: false
input_column: input
jit_mode_eval: false
label_names: null
label_smoothing_factor: 0.0
learning_rate: 0.0002
length_column_name: length
liger_kernel_config: null
load_best_model_at_end: false
local_rank: 0
log_dir: results/logs
log_file: results/logs/log.txt
log_level: passive
log_level_replica: warning
log_on_each_node: true
log_verbose: false
logger: !!python/object/apply:logging.getLogger
- LLM-Utils
logging_dir: /project2/jonmay_1426/ashokd/RAU/models/rwku/Llama-3.1-8B-Instruct_rt_forget-full/runs/Nov10_14-01-23_a11-01.hpc.usc.edu
logging_first_step: false
logging_nan_inf_filter: true
logging_steps: 250
logging_strategy: !!python/object/apply:transformers.trainer_utils.IntervalStrategy
- steps
lora_alpha: 16
lora_dropout: 0.05
lora_r: 8
lora_target_modules: null
loss_type: nll
lr_scheduler_kwargs: {}
lr_scheduler_type: !!python/object/apply:transformers.trainer_utils.SchedulerType
- cosine
max_grad_norm: 1.0
max_input_length: 512
max_length: 1024
max_steps: -1
max_test_samples: null
max_train_samples: null
max_valid_samples: null
metric_for_best_model: null
modality: lm
model_dir: /project2/jonmay_1426/ashokd/RAU/llm-utils/models
model_dtype: float16
model_init_kwargs: null
model_name: meta-llama/Llama-3.1-8B-Instruct
mp_parameters: ''
n_eval_output_batches: 1
neftune_noise_alpha: null
no_cuda: false
num_train_epochs: 1.0
num_workers: 4
openai_tmp_dir: ./openai_tmp_files
optim: !!python/object/apply:transformers.training_args.OptimizerNames
- adamw_torch
optim_args: null
optim_target_modules: null
output_column: chosen
output_dir: /project2/jonmay_1426/ashokd/RAU/models/rwku/Llama-3.1-8B-Instruct_rt_forget-full
overwrite_output_dir: false
packing: false
packing_strategy: bfd
pad_to_multiple_of: null
pad_token: null
padding_free: false
parallelism_config: null
past_index: -1
per_device_eval_batch_size: 4
per_device_train_batch_size: 4
per_gpu_eval_batch_size: null
per_gpu_train_batch_size: null
prediction_loss_only: false
pretrain_with_output: false
project: huggingface
project_root: /home1/ashokd/projects/RAU/llm-utils
push_to_hub: true
push_to_hub_model_id: null
push_to_hub_organization: null
push_to_hub_token: null
random_seed: null
ray_scope: last
rejected_column: null
remove_unused_columns: true
report_to: wandb
restore_callback_states_from_checkpoint: false
results_dir: results
resume_from_checkpoint: false
run_name: Llama-3.1-8B-Instruct_rt_forget-full
run_start_time: 2025-11-10-H22-M01-S22
save_on_each_node: false
save_only_model: false
save_safetensors: true
save_steps: 500
save_strategy: !!python/object/apply:transformers.trainer_utils.SaveStrategy
- steps
save_total_limit: 2
seed: 42
shuffle_buffer: 5000
skip_memory_metrics: true
storage_dir: /project2/jonmay_1426/ashokd/RAU/llm-utils
streaming: true
test_file: null
tf32: null
tmp_dir: /project2/jonmay_1426/ashokd/RAU/llm-utils/tmp
torch_compile: false
torch_compile_backend: null
torch_compile_mode: null
torch_empty_cache_steps: null
torchdynamo: null
tpu_metrics_debug: false
tpu_num_cores: null
trackio_space_id: trackio
train_file: /project2/jonmay_1426/ashokd/RAU/data/rwku/unlearn/forget_training.csv
train_validation_split: 0.9
training_kind: sft
use_bnb: false
use_cpu: false
use_legacy_prediction_loop: false
use_liger_kernel: false
use_mps_device: false
use_peft: true
validation_file: null
validation_test_split: null
warmup_ratio: 0.1
warmup_steps: 0
weight_decay: 0.01