Hugging Face's logo

Datasets:
Maple222
/
llmtcl

Modalities:
Text
ArXiv:
Dataset card Data Studio Files
xet
 Community
1
llmtcl
File size: 3,012 Bytes
f9dff7a
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
#!/bin/bash

MONTHS=(2407 2408 2409 2410 2411 2412 2501 2502 2503 2504 2505 2506)

# tinyllama 1b 3 epoch
#MAX_TOKENS=(
#   $((211812352*3))
#   $((177209344*3))
#   $((240123904*3))
#   $((258998272*3))
#   $((250609664*3))
#   $((304087040*3))
#   $((266338304*3))
#   $((330301440*3))
#   $((394264576*3))
#   $((375390208*3))
#   $((470810624*3))
#   $((508559360*3))
#)

# tinyllama 1b
MAX_TOKENS=(
   211812352
   177209344
   240123904
   258998272
   250609664
   304087040
   266338304
   330301440
   394264576
   375390208
   470810624
   508559360
)
# Qwen2-7B
# MAX_TOKENS=(
#  200278016
#  167772160
#  226492416
#  244318208
#  236978176
#  286261248
#  251658240
#  311427072
#  371195904
#  353370112
#  442499072
#  478150656
#)

# Qwen2-7B
# MAX_TOKENS=(
#  $((200278016*3))
#  $((167772160*3))
#  $((226492416*3))
#  $((244318208*3))
#  $((236978176*3))
#  $((286261248*3))
#  $((251658240*3))
#  $((311427072*3))
#  $((371195904*3))
#  $((353370112*3))
#  $((442499072*3))
#  $((478150656*3))
#)

# openllama 3b
#MAX_TOKENS=(
#  208666624
#  175112192
#  236978176
#  255852544
#  247463936
#  298844160
#  262144000
#  325058560
#  387973120
#  369098752
#  462422016
#  500170752
#)


OUT_ROOT="./out/pretrain/tinyllama_lr_plus"
#"./out/pretrain/qwen2_3_epoch"
#"./out/pretrain/tinyllama_3_epoch"
#"./out/pretrain/qwen2"
#"./out/pretrain/openllama"

LOG_ROOT="./out/pretrain/tinyllama_lr_plus/teelogs"
#"./out/pretrain/qwen2_3_epoch/teelogs"
#"./out/pretrain/tinyllama_3_epoch/teelogs"
#"./out/pretrain/qwen2/teelogs"
#"./out/pretrain/openllama/teelogs"

DATA_ROOT="litgpt/data/arxiv_tinyllama_tokenized/"
#"litgpt/data/arxiv_qwen2_tokenized/"
#
#"litgpt/data/arxiv_openllama_tokenized/"

CKPT_0="./checkpoints/TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T/"
#"./checkpoints/Qwen/Qwen2-7B/"
#"./checkpoints/TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T/"
#"./checkpoints/openlm-research/open_llama_3b/"

CONFIG_FILE="config_hub/pretrain/tinyllama_cl.yaml"
#"config_hub/pretrain/qwen2_7b_cl.yaml"
#"config_hub/pretrain/tinyllama_cl.yaml"
#"config_hub/pretrain/openllama_cl.yaml"


export CUDA_VISIBLE_DEVICES="4,5,6,7"
mkdir -p "$LOG_ROOT"

START_IDX=0
for ((i=$START_IDX; i<${#MONTHS[@]}; i++)); do
    MONTH=${MONTHS[$i]}
    TOKENS=${MAX_TOKENS[$i]}

    if [ $i -eq 0 ]; then
        INIT_CKPT="$CKPT_0"
    else
        PREV_MONTH=${MONTHS[$((i-1))]}
        INIT_CKPT="${OUT_ROOT}/${PREV_MONTH}/final"
    fi

    OUT_DIR="${OUT_ROOT}/${MONTH}"
    LOG_FILE="${LOG_ROOT}/${MONTH}.txt"
    DATA_DIR="${DATA_ROOT}/${MONTH}/"

    mkdir -p "$OUT_DIR"

    echo "===== Start Training $MONTH ====="
    echo "init_ckpt: $INIT_CKPT"
    echo "out_dir: $OUT_DIR"
    echo "max_tokens: $TOKENS"

    litgpt pretrain \
        --config $CONFIG_FILE \
        --train.max_tokens $TOKENS \
        --out_dir $OUT_DIR \
        --initial_checkpoint_dir $INIT_CKPT \
        --data_dir $DATA_DIR \
        2>&1 | tee "$LOG_FILE"

    echo "===== $MONTH Ended ====="
    sleep 60
done