spamassassin无学习数据，是无法自动学习？

scpe

最近英文垃圾邮件好多，我的spamassassin开启了bayes自动学习，可怎么一直没有学习数据？看有的邮件里面的autolearn=spam或autolearn=ham，是否系统又在学习。我已将学习邮件数调整为20。不知道什么原因，请大家帮我看看吧。谢谢！
[root@dypc spamassassin]# sa-learn --dump magic
0.000          0          3          0  non-token data: bayes db version
0.000          0          0          0  non-token data: nspam
0.000          0          0          0  non-token data: nham
0.000          0          0          0  non-token data: ntokens
0.000          0          0          0  non-token data: oldest atime
0.000          0          0          0  non-token data: newest atime
0.000          0          0          0  non-token data: last journal sync atime
0.000          0          0          0  non-token data: last expiry atime
0.000          0          0          0  non-token data: last expire atime delta
0.000          0          0          0  non-token data: last expire reduction count

以下是一封垃圾邮件的邮件头：

Subject: ****SPAM**** Replica Watches
Date: Sun, 11 Feb 2007 21:21:53 +0300
Message-Id: <40f501c74e09$a2a38ea1$a4ecfa86@backpackers.net>
X-Spam-Flag: YES
X-Spam-Checker-Version: SpamAssassin 3.0.2 (2004-11-16) on dypc.net
X-Spam-Level: ***********
X-Spam-Status: Yes, score=12.0 required=5.0 tests=BAYES_50,HTML_FONT_BIG,
        HTML_MESSAGE,RCVD_IN_BL_SPAMCOP_NET,RCVD_IN_XBL,URIBL_AB_SURBL,
        URIBL_SBL,URIBL_SC_SURBL autolearn=spam version=3.0.2
MIME-Version: 1.0
Content-Type: multipart/mixed; boundary="----------=_45CF5BC1.78C35753"

以下是我的local.cf配置文件内容：

# These values can be overridden by editing ~/.spamassassin/user_prefs.cf
# (see spamassassin(1) for details)

# These should be safe assumptions and allow for simple visual sifting
# without risking lost emails.

#required_hits 5
required_score 5.0
rewrite_header Subject ****SPAM****
report_safe 1
use_bayes 1
#auto_learn 1
bayes_auto_learn 1
bayes_min_spam_num 20
bayes_min_ham_num 20
skip_rbl_checks 0
ok_languages zh en
ok_locales zh en
score HEAD_ILLEGAL_CHARS 0
score SUBJ_ILLEGAL_CHARS 0
score DATE_IN_PAST_03_06 0
score UPPERCASE_25_50 0
score UPPERCASE_50_75 0
score UPPERCASE_75_100 0
score X_MSMAIL_PRIORITY_HIGH 0
score X_PRIORITY_HIGH 0
score TO_TXT 100
score RATWARE_HASH_2 100
score RATWARE_HASH_2_V2 100
score BAYES_99 0.1
score BAYES_80 0.1
score BAYES_60 0.1
score FROM_ILLEGAL_CHARS 0.1
score MIME_BASE64_TEXT 0.1
score NO_RDNS_DOTCOM_HELO 0.1
score CHINA_HEADER 0.1
score NO_REAL_NAME 0.2
score HTML_MESSAGE 0.2
score MIME_HTML_ONLY 0.2
score MIME_HTML_ONLY_MULTI 0.2
score FORGED_MUA_OUTLOOK 0.2
score FORGED_HOTMAIL_RCVD 0.2
score FORGED_OUTLOOK_TAGS 0.2
score MAILTO_TO_SPAM_ADDR 0.2
###################################
# use china anti-spam CBL/CDL     #
# URL:http://www.anti-spam.org.cn #
###################################
header RCVD_IN_CBL eval:check_rbl('cblplus','cblplus.anti-spam.org.cn.')
describe RCVD_IN_CBL Received via a relay in cbl.anti-spam.org.cn
tflags RCVD_IN_CBL net

header RCVD_IN_CDL eval:check_rbl('cdl-notfirsthop','cdlplus.ant-spam.org.cn')
describe RCVD_IN_CDL CDL:dialup sender did non-local SMTP
tflags RCVD_IN_CDL net

#SCORE
score RCVD_IN_CBL 4.0
score RCVD_IN_CDL 3.0