video caption based on liunx make

目前在做video caption 在用liunx跑的代碼，裏面用到了個make 命令 ** “ make -f Makefile_msrvtt_svo test GID=0 EXP_NAME=xe FEATS="irv2 c3d category" BFEATS="roi_feat roi_box" USE_RL=0 CST=0 USE_MIXER=0 SCB_CAPTIONS=0 LOGLEVEL=DEBUG LAMBDA=20”， 然後我按照命令去做，結果報錯了 ** “make: *** 没有规则可制作目标“output/metadata/msrvtt_train_sequencelabel.h5”，由“output/model_svo/xe/irv2c3dcategory_msrvtt_concat_CIDEr_64_0.0001_20.pth” 需求。 停止。”，

img

img

因爲之前也沒學過make,所以一直搞不定，求各位前輩幫幫忙！

你把运行日志或者报错复制出来发给我

这个就是Makefile


这个就是那个makefile
### Directory Setting
IN_DIR=input
OUT_DIR=output
META_DIR=$(OUT_DIR)/metadata
FEAT_DIR=$(OUT_DIR)/feature
MODEL_DIR=$(OUT_DIR)/model_svo

MSRVTT2016_DIR=$(IN_DIR)/msrvtt
MSVD_DIR=$(IN_DIR)/msvd
###MSRVTT2017_DIR=$(IN_DIR)/msrvtt2017
###YT2T_DIR=$(IN_DIR)/yt2t

SPLITS=train val test
DATASETS=msrvtt#  msrvtt2016 msvd
###DATASETS=msrvtt# yt2t msrvtt2017 tvvtt

WORD_COUNT_THRESHOLD?=0  # in output/metadata this threshold was 0; was 3 in output/metadata2017
MAX_SEQ_LEN?=20          # in output/metadata seqlen was 20; was 30 in output/metadata2017

GID?=1

DATASET?=msrvtt
TRAIN_DATASET?=$(DATASET)
VAL_DATASET?=$(DATASET)
TEST_DATASET?=$(DATASET)
TRAIN_SPLIT?=train
VAL_SPLIT?=val
TEST_SPLIT?=test

LEARNING_RATE?=0.0001
LR_UPDATE?=200
BATCH_SIZE?=64
TRAIN_SEQ_PER_IMG?=20
TEST_SEQ_PER_IMG?=20
RNN_SIZE?=512

PRINT_INTERVAL?=20
MAX_PATIENCE?=50 # FOR EARLY STOPPING
SAVE_CHECKPOINT_FROM?=1

MAX_EPOCHS?=200
NUM_CHUNKS?=1
PRINT_ATT_COEF?=0
BEAM_SIZE?=5
RNN_TYPE?=lstm
INPUT_ENCODING_SIZE?=512

TODAY=20221118
EXP_NAME?=exp_$(DATASET)_$(TODAY)
VAL_LANG_EVAL?=1
TEST_LANG_EVAL?=1
EVAL_METRIC?=CIDEr
START_FROM?=No
MODEL_TYPE?=concat
POOLING?=mp
CAT_TYPE=glove
LOGLEVEL?=INFO

SS_MAX_PROB?=0.25
USE_CST?=0
SCB_CAPTIONS?=20
SCB_BASELINE?=1
USE_RL?=0
USE_RL_AFTER?=0
USE_EOS?=0
USE_MIXER?=0
MIXER_FROM?=-1
SS_K?=100
LAMBDA?=10

FEAT1?=resnet
FEAT2?=c3d
FEAT3?=mfcc
FEAT4?=category

FEATS?=$(FEAT1) $(FEAT2) $(FEAT3) $(FEAT4)

BFEAT1?=roi_feat
BFEAT2?=roi_box

BFEAT?=$(BFEAT1) $(BFEAT2)

TRAIN_ID=$(TRAIN_DATASET)_$(MODEL_TYPE)_$(EVAL_METRIC)_$(BATCH_SIZE)_$(LEARNING_RATE)_$(LAMBDA)


#####################################################################################################################

noop=
space=$(noop) $(noop)

TRAIN_OPT=--beam_size $(BEAM_SIZE) --labda $(LAMBDA) --rnn_type $(RNN_TYPE) --input_encoding_size $(INPUT_ENCODING_SIZE) --max_patience $(MAX_PATIENCE) --eval_metric $(EVAL_METRIC) --print_log_interval $(PRINT_INTERVAL)\
    --language_eval $(VAL_LANG_EVAL) --max_epochs $(MAX_EPOCHS) --rnn_size $(RNN_SIZE) \
    --train_seq_per_img $(TRAIN_SEQ_PER_IMG) --test_seq_per_img $(TEST_SEQ_PER_IMG) \
    --batch_size $(BATCH_SIZE) --test_batch_size $(BATCH_SIZE) --learning_rate $(LEARNING_RATE) --lr_update $(LR_UPDATE) \
    --save_checkpoint_from $(SAVE_CHECKPOINT_FROM) --num_chunks $(NUM_CHUNKS) \
    --train_cached_tokens $(META_DIR)/$(TRAIN_DATASET)_train_ciderdf.pkl \
    --ss_k $(SS_K) --use_rl_after $(USE_RL_AFTER) --ss_max_prob $(SS_MAX_PROB) \
    --use_rl $(USE_RL) --use_mixer $(USE_MIXER) --mixer_from $(MIXER_FROM) \
    --use_cst $(USE_CST) --scb_captions $(SCB_CAPTIONS) --scb_baseline $(SCB_BASELINE) \
    --loglevel $(LOGLEVEL) --model_type $(MODEL_TYPE) --use_eos $(USE_EOS) \
    --model_file $@ --start_from $(START_FROM) --result_file $(basename $@)_test.json \
    2>&1 | tee $(basename $@).log

TEST_OPT=--beam_size $(BEAM_SIZE) \
    --rnn_type $(RNN_TYPE) \
    --input_encoding_size $(INPUT_ENCODING_SIZE) \
    --language_eval $(VAL_LANG_EVAL) \
    --test_seq_per_img $(TEST_SEQ_PER_IMG) \
    --test_batch_size $(BATCH_SIZE) \
    --loglevel $(LOGLEVEL) \
    --result_file $@

train: $(MODEL_DIR)/$(EXP_NAME)/$(subst $(space),$(noop),$(FEATS))_$(TRAIN_ID).pth
$(MODEL_DIR)/$(EXP_NAME)/$(subst $(space),$(noop),$(FEATS))_$(TRAIN_ID).pth: \
    $(META_DIR)/$(TRAIN_DATASET)_$(TRAIN_SPLIT)_sequencelabel.h5 \
    $(META_DIR)/$(VAL_DATASET)_$(VAL_SPLIT)_sequencelabel.h5 \
    $(META_DIR)/$(TEST_DATASET)_$(TEST_SPLIT)_sequencelabel.h5 \
    $(META_DIR)/$(TRAIN_DATASET)_$(TRAIN_SPLIT)_cocofmt.json \
    $(META_DIR)/$(VAL_DATASET)_$(VAL_SPLIT)_cocofmt.json \
    $(META_DIR)/$(TEST_DATASET)_$(TEST_SPLIT)_cocofmt.json \
    $(META_DIR)/$(TRAIN_DATASET)_$(TRAIN_SPLIT)_evalscores.pkl \
    $(patsubst %,$(FEAT_DIR)/$(TRAIN_DATASET)_$(TRAIN_SPLIT)_%_mp$(NUM_CHUNKS).h5,$(FEATS)) \
    $(patsubst %,$(FEAT_DIR)/$(VAL_DATASET)_$(VAL_SPLIT)_%_mp$(NUM_CHUNKS).h5,$(FEATS)) \
    $(patsubst %,$(FEAT_DIR)/$(TEST_DATASET)_$(TEST_SPLIT)_%_mp$(NUM_CHUNKS).h5,$(FEATS)) \
    $(patsubst %,$(FEAT_DIR)/$(TRAIN_DATASET)_%.h5,$(BFEATS)) 

    mkdir -p $(MODEL_DIR)/$(EXP_NAME)
    CUDA_VISIBLE_DEVICES=$(GID) python train_svo.py \
        --train_label_h5 $(word 1,$^) \
        --val_label_h5 $(word 2,$^) \
        --test_label_h5 $(word 3,$^) \
        --train_cocofmt_file $(word 4,$^) \
        --val_cocofmt_file $(word 5,$^) \
        --test_cocofmt_file $(word 6,$^) \
        --train_bcmrscores_pkl $(word 7,$^) \
        --train_feat_h5 $(patsubst %,$(FEAT_DIR)/$(TRAIN_DATASET)_$(TRAIN_SPLIT)_%_mp$(NUM_CHUNKS).h5,$(FEATS))\
        --val_feat_h5 $(patsubst %,$(FEAT_DIR)/$(VAL_DATASET)_$(VAL_SPLIT)_%_mp$(NUM_CHUNKS).h5,$(FEATS))\
        --test_feat_h5 $(patsubst %,$(FEAT_DIR)/$(TEST_DATASET)_$(TEST_SPLIT)_%_mp$(NUM_CHUNKS).h5,$(FEATS))\
        --bfeat_h5 $(patsubst %,$(FEAT_DIR)/$(TRAIN_DATASET)_%.h5,$(BFEATS)) \
        --fr_size_h5 $(FEAT_DIR)/$(TRAIN_DATASET)_fr_size.h5 \
        $(TRAIN_OPT)

test: $(MODEL_DIR)/$(EXP_NAME)/$(subst $(space),$(noop),$(FEATS))_$(TRAIN_ID)_test.json
$(MODEL_DIR)/$(EXP_NAME)/$(subst $(space),$(noop),$(FEATS))_$(TRAIN_ID)_test.json: \
    $(MODEL_DIR)/$(EXP_NAME)/$(subst $(space),$(noop),$(FEATS))_$(TRAIN_ID).pth \
    $(META_DIR)/$(TEST_DATASET)_$(TEST_SPLIT)_sequencelabel.h5 \
    $(META_DIR)/$(TEST_DATASET)_$(TEST_SPLIT)_cocofmt.json \
    $(patsubst %,$(FEAT_DIR)/$(TEST_DATASET)_$(TEST_SPLIT)_%_mp$(NUM_CHUNKS).h5,$(FEATS))
    CUDA_VISIBLE_DEVICES=$(GID) python test_svo.py \
        --model_file $(word 1,$^) \
        --test_label_h5 $(word 2,$^) \
        --test_cocofmt_file $(word 3,$^) \
        --test_feat_h5 $(patsubst %,$(FEAT_DIR)/$(TEST_DATASET)_$(TEST_SPLIT)_%_mp$(NUM_CHUNKS).h5,$(FEATS))\
        --bfeat_h5 $(patsubst %,$(FEAT_DIR)/$(TRAIN_DATASET)_%.h5,$(BFEATS)) \
        --fr_size_h5 $(FEAT_DIR)/$(TRAIN_DATASET)_fr_size.h5 \
        $(TEST_OPT)


# You can use the wildcard with .PRECIOUS.
.PRECIOUS: %.pth

# If you want all intermediates to remain
.SECONDARY: