Delete data/configs_hc.yml
Browse files- data/configs_hc.yml +0 -91
data/configs_hc.yml
DELETED
@@ -1,91 +0,0 @@
|
|
1 |
-
base_model: mistralai/Mistral-7B-v0.1
|
2 |
-
model_type: MistralForCausalLM
|
3 |
-
tokenizer_type: LlamaTokenizer
|
4 |
-
is_mistral_derived_model: true
|
5 |
-
|
6 |
-
load_in_8bit: false
|
7 |
-
load_in_4bit: true
|
8 |
-
strict: false
|
9 |
-
|
10 |
-
lora_fan_in_fan_out: false
|
11 |
-
data_seed: 49
|
12 |
-
seed: 49
|
13 |
-
|
14 |
-
datasets:
|
15 |
-
- path: /teamspace/studios/this_studio/axolotl/finetune_maven/hc-mistral-alpaca/data/alpaca_synth_queries_healed.jsonl
|
16 |
-
type: sharegpt
|
17 |
-
conversation: alpaca
|
18 |
-
dataset_prepared_path: last_run_prepared
|
19 |
-
val_set_size: 0.1
|
20 |
-
output_dir: ./qlora-alpaca-out
|
21 |
-
hub_model_id: valLabsz/clssfcn-naming
|
22 |
-
#hub_model_id: hamel/hc-mistral-alpaca
|
23 |
-
|
24 |
-
|
25 |
-
adapter: qlora
|
26 |
-
lora_model_dir:
|
27 |
-
|
28 |
-
sequence_len: 896
|
29 |
-
sample_packing: false
|
30 |
-
pad_to_sequence_len: true
|
31 |
-
|
32 |
-
lora_r: 32
|
33 |
-
lora_alpha: 16
|
34 |
-
lora_dropout: 0.05
|
35 |
-
lora_target_linear: true
|
36 |
-
lora_fan_in_fan_out:
|
37 |
-
lora_target_modules:
|
38 |
-
- gate_proj
|
39 |
-
- down_proj
|
40 |
-
- up_proj
|
41 |
-
- q_proj
|
42 |
-
- v_proj
|
43 |
-
- k_proj
|
44 |
-
- o_proj
|
45 |
-
|
46 |
-
#wandb_project: hc-axolotl-mistral
|
47 |
-
#wandb_entity: hamelsmu
|
48 |
-
|
49 |
-
gradient_accumulation_steps: 4
|
50 |
-
micro_batch_size: 16
|
51 |
-
eval_batch_size: 16
|
52 |
-
num_epochs: 3
|
53 |
-
optimizer: adamw_bnb_8bit
|
54 |
-
lr_scheduler: cosine
|
55 |
-
learning_rate: 0.0002
|
56 |
-
max_grad_norm: 1.0
|
57 |
-
adam_beta2: 0.95
|
58 |
-
adam_epsilon: 0.00001
|
59 |
-
save_total_limit: 12
|
60 |
-
|
61 |
-
train_on_inputs: false
|
62 |
-
group_by_length: false
|
63 |
-
bf16: true
|
64 |
-
fp16: false
|
65 |
-
tf32: false
|
66 |
-
|
67 |
-
gradient_checkpointing: true
|
68 |
-
early_stopping_patience:
|
69 |
-
resume_from_checkpoint:
|
70 |
-
local_rank:
|
71 |
-
logging_steps: 1
|
72 |
-
xformers_attention:
|
73 |
-
flash_attention: true
|
74 |
-
|
75 |
-
loss_watchdog_threshold: 5.0
|
76 |
-
loss_watchdog_patience: 3
|
77 |
-
|
78 |
-
warmup_steps: 20
|
79 |
-
evals_per_epoch: 4
|
80 |
-
eval_table_size:
|
81 |
-
eval_table_max_new_tokens: 128
|
82 |
-
saves_per_epoch: 6
|
83 |
-
debug:
|
84 |
-
weight_decay: 0.0
|
85 |
-
fsdp:
|
86 |
-
fsdp_config:
|
87 |
-
special_tokens:
|
88 |
-
bos_token: "<s>"
|
89 |
-
eos_token: "</s>"
|
90 |
-
unk_token: "<unk>"
|
91 |
-
save_safetensors: true
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|