Upload llama-1B/8_GPUS/dp-2_tp-1_pp-4_mbz-16
Browse files
llama-1B/8_GPUS/dp-2_tp-1_pp-4_mbz-16/bench.slurm
ADDED
|
@@ -0,0 +1,111 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
#!/bin/bash
|
| 2 |
+
|
| 3 |
+
#SBATCH --job-name=bench_cluster
|
| 4 |
+
#SBATCH --time=02:00:00
|
| 5 |
+
#SBATCH --partition=hopper-prod
|
| 6 |
+
#SBATCH --nodes=1
|
| 7 |
+
#SBATCH --gres=gpu:8
|
| 8 |
+
#SBATCH --qos=normal
|
| 9 |
+
#SBATCH --ntasks-per-node=1
|
| 10 |
+
#SBATCH --cpus-per-task=96
|
| 11 |
+
#SBATCH --exclusive
|
| 12 |
+
#SBATCH --output=/fsx/ferdinandmom/ferdinand-hf/bench_cluster/results/llama-1B/8_GPUS/dp-2_tp-1_pp-4_mbz-16/log.out
|
| 13 |
+
#SBATCH --error=/fsx/ferdinandmom/ferdinand-hf/bench_cluster/results/llama-1B/8_GPUS/dp-2_tp-1_pp-4_mbz-16/log.out
|
| 14 |
+
|
| 15 |
+
# Function to update status based on squeue output
|
| 16 |
+
update_status() {
|
| 17 |
+
job_id=$1
|
| 18 |
+
status_file=$2
|
| 19 |
+
# For unknown reasons, it doenst update status for pending. It only works for running
|
| 20 |
+
while true; do
|
| 21 |
+
job_status=$(squeue --job $job_id --noheader --format=%T)
|
| 22 |
+
echo "Job status: $job_status"
|
| 23 |
+
if [ -z "$job_status" ]; then
|
| 24 |
+
# Job has finished or is not found
|
| 25 |
+
break
|
| 26 |
+
elif [ "$job_status" = "RUNNING" ]; then
|
| 27 |
+
printf "running" > $status_file
|
| 28 |
+
break
|
| 29 |
+
fi
|
| 30 |
+
sleep 10
|
| 31 |
+
done
|
| 32 |
+
}
|
| 33 |
+
|
| 34 |
+
# Misc initializations.
|
| 35 |
+
echo "========================"
|
| 36 |
+
echo "START TIME: $(date)"
|
| 37 |
+
source /fsx/ferdinandmom/miniforge3/etc/profile.d/conda.sh
|
| 38 |
+
conda activate /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster
|
| 39 |
+
echo python3 version = $(python3 --version)
|
| 40 |
+
echo "========================"
|
| 41 |
+
|
| 42 |
+
# Slurm stuff
|
| 43 |
+
export HOSTNAMES=$(scontrol show hostnames "$SLURM_JOB_NODELIST")
|
| 44 |
+
export MASTER_ADDR=$(scontrol show hostnames "$SLURM_JOB_NODELIST" | head -n 1)
|
| 45 |
+
export MASTER_PORT=$((1024 + RANDOM % 64511))
|
| 46 |
+
|
| 47 |
+
export TMPDIR=/scratch
|
| 48 |
+
export HF_DATASETS_CACHE="/admin/home/ferdinand_mom/.cache"
|
| 49 |
+
export CUBLAS_WORKSPACE_CONFIG=":4096:8"
|
| 50 |
+
export CUDA_DEVICE_MAX_CONNECTIONS="1"
|
| 51 |
+
|
| 52 |
+
huggingface-cli login --token $HUGGINGFACE_TOKEN
|
| 53 |
+
|
| 54 |
+
|
| 55 |
+
NANOTRON_REPO="/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron"
|
| 56 |
+
CMD="$NANOTRON_REPO/run_train.py --config-file /fsx/ferdinandmom/ferdinand-hf/bench_cluster/results/llama-1B/8_GPUS/dp-2_tp-1_pp-4_mbz-16/config.yaml"
|
| 57 |
+
|
| 58 |
+
LAUNCHER="torchrun \
|
| 59 |
+
--nproc_per_node 8 \
|
| 60 |
+
--nnodes 1 \
|
| 61 |
+
--rdzv_endpoint ${MASTER_ADDR}:${MASTER_PORT} \
|
| 62 |
+
--rdzv_backend c10d \
|
| 63 |
+
--max_restarts 0 \
|
| 64 |
+
--tee 3 \
|
| 65 |
+
--node_rank ${SLURM_PROCID}"
|
| 66 |
+
|
| 67 |
+
# Checkout the bench_cluster branch
|
| 68 |
+
cd $NANOTRON_REPO
|
| 69 |
+
git checkout bench_cluster
|
| 70 |
+
cd ..
|
| 71 |
+
# Get the current job ID
|
| 72 |
+
job_id=${SLURM_JOB_ID}
|
| 73 |
+
|
| 74 |
+
# Update status to "pending" or "running" in the background
|
| 75 |
+
update_status $job_id /fsx/ferdinandmom/ferdinand-hf/bench_cluster/results/llama-1B/8_GPUS/dp-2_tp-1_pp-4_mbz-16/status.txt &
|
| 76 |
+
|
| 77 |
+
# Run the main command
|
| 78 |
+
srun -u $LAUNCHER $CMD
|
| 79 |
+
exit_status=$?
|
| 80 |
+
|
| 81 |
+
# Update status based on the exit status of `srun`
|
| 82 |
+
if [ $exit_status -eq 0 ]; then
|
| 83 |
+
printf "completed" > /fsx/ferdinandmom/ferdinand-hf/bench_cluster/results/llama-1B/8_GPUS/dp-2_tp-1_pp-4_mbz-16/status.txt
|
| 84 |
+
else
|
| 85 |
+
if grep -q "OutOfMemoryError" /fsx/ferdinandmom/ferdinand-hf/bench_cluster/results/llama-1B/8_GPUS/dp-2_tp-1_pp-4_mbz-16/log.out; then
|
| 86 |
+
printf "oom" > /fsx/ferdinandmom/ferdinand-hf/bench_cluster/results/llama-1B/8_GPUS/dp-2_tp-1_pp-4_mbz-16/status.txt
|
| 87 |
+
elif grep -q " CUDA error: an illegal memory access" /fsx/ferdinandmom/ferdinand-hf/bench_cluster/results/llama-1B/8_GPUS/dp-2_tp-1_pp-4_mbz-16/log.out; then
|
| 88 |
+
printf "oom" > /fsx/ferdinandmom/ferdinand-hf/bench_cluster/results/llama-1B/8_GPUS/dp-2_tp-1_pp-4_mbz-16/status.txt
|
| 89 |
+
elif grep -q "Timeout at NCCL" /fsx/ferdinandmom/ferdinand-hf/bench_cluster/results/llama-1B/8_GPUS/dp-2_tp-1_pp-4_mbz-16/log.out; then
|
| 90 |
+
printf "timeout" > /fsx/ferdinandmom/ferdinand-hf/bench_cluster/results/llama-1B/8_GPUS/dp-2_tp-1_pp-4_mbz-16/status.txt
|
| 91 |
+
else
|
| 92 |
+
printf "fail" > /fsx/ferdinandmom/ferdinand-hf/bench_cluster/results/llama-1B/8_GPUS/dp-2_tp-1_pp-4_mbz-16/status.txt
|
| 93 |
+
fi
|
| 94 |
+
fi
|
| 95 |
+
|
| 96 |
+
# Run the report script if the job completed successfully
|
| 97 |
+
if [ $exit_status -eq 0 ]; then
|
| 98 |
+
python /fsx/ferdinandmom/ferdinand-hf/bench_cluster/main.py report --inp_dir /fsx/ferdinandmom/ferdinand-hf/bench_cluster/results/llama-1B/8_GPUS/dp-2_tp-1_pp-4_mbz-16 --is_logs
|
| 99 |
+
python /fsx/ferdinandmom/ferdinand-hf/bench_cluster/main.py report --inp_dir /fsx/ferdinandmom/ferdinand-hf/bench_cluster/results/llama-1B/8_GPUS/dp-2_tp-1_pp-4_mbz-16 --is_profiler
|
| 100 |
+
fi
|
| 101 |
+
|
| 102 |
+
|
| 103 |
+
# Push to hub the folder using huggingface_cli
|
| 104 |
+
huggingface-cli upload nanotron/bench_cluster /fsx/ferdinandmom/ferdinand-hf/bench_cluster/results/llama-1B/8_GPUS/dp-2_tp-1_pp-4_mbz-16 llama-1B/8_GPUS/dp-2_tp-1_pp-4_mbz-16 --commit-message "Upload llama-1B/8_GPUS/dp-2_tp-1_pp-4_mbz-16"
|
| 105 |
+
|
| 106 |
+
# Verify the upload
|
| 107 |
+
if [ $? -eq 0 ]; then
|
| 108 |
+
echo "Uploading to Huggingface Hub successful"
|
| 109 |
+
else
|
| 110 |
+
echo "Failed to upload to Huggingface Hub"
|
| 111 |
+
fi
|
llama-1B/8_GPUS/dp-2_tp-1_pp-4_mbz-16/config.yaml
ADDED
|
@@ -0,0 +1,90 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
general:
|
| 2 |
+
project: bench_cluster
|
| 3 |
+
seed: 42
|
| 4 |
+
model:
|
| 5 |
+
ddp_bucket_cap_mb: 25
|
| 6 |
+
dtype: bfloat16
|
| 7 |
+
init_method:
|
| 8 |
+
std: 0.025
|
| 9 |
+
make_vocab_size_divisible_by: 1
|
| 10 |
+
model_config:
|
| 11 |
+
bos_token_id: 1
|
| 12 |
+
eos_token_id: 2
|
| 13 |
+
hidden_act: silu
|
| 14 |
+
hidden_size: 2048
|
| 15 |
+
initializer_range: 0.02
|
| 16 |
+
intermediate_size: 4096
|
| 17 |
+
is_llama_config: true
|
| 18 |
+
max_position_embeddings: 4096
|
| 19 |
+
num_attention_heads: 32
|
| 20 |
+
num_hidden_layers: 24
|
| 21 |
+
num_key_value_heads: 32
|
| 22 |
+
pad_token_id: null
|
| 23 |
+
pretraining_tp: 1
|
| 24 |
+
rms_norm_eps: 1.0e-05
|
| 25 |
+
rope_scaling: null
|
| 26 |
+
rope_theta: 10000.0
|
| 27 |
+
tie_word_embeddings: true
|
| 28 |
+
use_cache: true
|
| 29 |
+
vocab_size: 50257
|
| 30 |
+
optimizer:
|
| 31 |
+
accumulate_grad_in_fp32: true
|
| 32 |
+
clip_grad: 1.0
|
| 33 |
+
learning_rate_scheduler:
|
| 34 |
+
learning_rate: 0.0001
|
| 35 |
+
lr_decay_style: linear
|
| 36 |
+
lr_warmup_style: linear
|
| 37 |
+
lr_warmup_steps: 1
|
| 38 |
+
min_decay_lr: 1.0e-05
|
| 39 |
+
optimizer_factory:
|
| 40 |
+
adam_beta1: 0.9
|
| 41 |
+
adam_beta2: 0.95
|
| 42 |
+
adam_eps: 1.0e-08
|
| 43 |
+
name: adamW
|
| 44 |
+
torch_adam_is_fused: true
|
| 45 |
+
weight_decay: 0.01
|
| 46 |
+
zero_stage: 1
|
| 47 |
+
parallelism:
|
| 48 |
+
dp: 2
|
| 49 |
+
expert_parallel_size: 1
|
| 50 |
+
pp: 4
|
| 51 |
+
pp_engine: 1f1b
|
| 52 |
+
tp: 1
|
| 53 |
+
tp_linear_async_communication: false
|
| 54 |
+
tp_mode: REDUCE_SCATTER
|
| 55 |
+
profiler:
|
| 56 |
+
profiler_export_path: /fsx/ferdinandmom/ferdinand-hf/bench_cluster/results/llama-1B/8_GPUS/dp-2_tp-1_pp-4_mbz-16
|
| 57 |
+
tokenizer:
|
| 58 |
+
tokenizer_max_length: null
|
| 59 |
+
tokenizer_name_or_path: openai-community/gpt2
|
| 60 |
+
tokenizer_revision: null
|
| 61 |
+
data_stages:
|
| 62 |
+
- name: Training Stage
|
| 63 |
+
start_training_step: 1
|
| 64 |
+
data:
|
| 65 |
+
dataset:
|
| 66 |
+
dataset_overwrite_cache: false
|
| 67 |
+
dataset_processing_num_proc_per_process: 64
|
| 68 |
+
hf_dataset_config_name: null
|
| 69 |
+
hf_dataset_or_datasets: roneneldan/TinyStories
|
| 70 |
+
hf_dataset_splits: train
|
| 71 |
+
text_column_name: text
|
| 72 |
+
num_loading_workers: 0
|
| 73 |
+
seed: 42
|
| 74 |
+
lighteval: null
|
| 75 |
+
tokens:
|
| 76 |
+
train_steps: 20
|
| 77 |
+
val_check_interval: -1
|
| 78 |
+
batch_accumulation_per_replica: 32
|
| 79 |
+
limit_test_batches: 0
|
| 80 |
+
limit_val_batches: 0
|
| 81 |
+
micro_batch_size: 16
|
| 82 |
+
sequence_length: 4096
|
| 83 |
+
logging:
|
| 84 |
+
iteration_step_info_interval: 1
|
| 85 |
+
log_level: info
|
| 86 |
+
log_level_replica: info
|
| 87 |
+
checkpoints:
|
| 88 |
+
checkpoint_interval: 100000
|
| 89 |
+
checkpoints_path: /dev/null
|
| 90 |
+
resume_checkpoint_path: null
|
llama-1B/8_GPUS/dp-2_tp-1_pp-4_mbz-16/log.out
ADDED
|
@@ -0,0 +1,663 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
========================
|
| 2 |
+
START TIME: Wed Jul 3 23:01:26 UTC 2024
|
| 3 |
+
python3 version = Python 3.10.14
|
| 4 |
+
========================
|
| 5 |
+
The token has not been saved to the git credentials helper. Pass `add_to_git_credential=True` in this function directly or `--add-to-git-credential` if using via `huggingface-cli` if you want to set the git credential as well.
|
| 6 |
+
Token is valid (permission: write).
|
| 7 |
+
Your token has been saved to /admin/home/ferdinand_mom/.cache/huggingface/token
|
| 8 |
+
Login successful
|
| 9 |
+
Already on 'bench_cluster'
|
| 10 |
+
M examples/config_tiny_llama.py
|
| 11 |
+
M examples/config_tiny_llama.yaml
|
| 12 |
+
M examples/train_tiny_llama.sh
|
| 13 |
+
M src/nanotron/models/llama.py
|
| 14 |
+
M src/nanotron/trainer.py
|
| 15 |
+
Your branch is up to date with 'origin/bench_cluster'.
|
| 16 |
+
Job status: RUNNING
|
| 17 |
+
W0703 23:01:29.951000 139658830735168 torch/distributed/run.py:757]
|
| 18 |
+
W0703 23:01:29.951000 139658830735168 torch/distributed/run.py:757] *****************************************
|
| 19 |
+
W0703 23:01:29.951000 139658830735168 torch/distributed/run.py:757] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed.
|
| 20 |
+
W0703 23:01:29.951000 139658830735168 torch/distributed/run.py:757] *****************************************
|
| 21 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: Config:
|
| 22 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: Config(general=GeneralArgs(project='bench_cluster',
|
| 23 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: run='%date_%jobid',
|
| 24 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: seed=42,
|
| 25 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: step=None,
|
| 26 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: consumed_train_samples=None,
|
| 27 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: benchmark_csv_path=None,
|
| 28 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: ignore_sanity_checks=True),
|
| 29 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: parallelism=ParallelismArgs(dp=2,
|
| 30 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: pp=4,
|
| 31 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: tp=1,
|
| 32 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7efd6ab78730>,
|
| 33 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
|
| 34 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: tp_linear_async_communication=False,
|
| 35 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: expert_parallel_size=1),
|
| 36 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: model=ModelArgs(model_config=LlamaConfig(bos_token_id=1,
|
| 37 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: eos_token_id=2,
|
| 38 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: hidden_act='silu',
|
| 39 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: hidden_size=2048,
|
| 40 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: initializer_range=0.02,
|
| 41 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: intermediate_size=4096,
|
| 42 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: is_llama_config=True,
|
| 43 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: max_position_embeddings=4096,
|
| 44 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: num_attention_heads=32,
|
| 45 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: num_hidden_layers=24,
|
| 46 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: num_key_value_heads=32,
|
| 47 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: pad_token_id=None,
|
| 48 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: pretraining_tp=1,
|
| 49 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: rms_norm_eps=1e-05,
|
| 50 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: rope_scaling=None,
|
| 51 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: rope_theta=10000.0,
|
| 52 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: tie_word_embeddings=True,
|
| 53 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: use_cache=True,
|
| 54 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: vocab_size=50257),
|
| 55 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: init_method=RandomInit(std=0.025),
|
| 56 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: dtype=torch.bfloat16,
|
| 57 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: make_vocab_size_divisible_by=1,
|
| 58 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: ddp_bucket_cap_mb=25),
|
| 59 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: tokenizer=TokenizerArgs(tokenizer_name_or_path='openai-community/gpt2',
|
| 60 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: tokenizer_revision=None,
|
| 61 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: tokenizer_max_length=None),
|
| 62 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: checkpoints=CheckpointsArgs(checkpoints_path=Path('/dev/null'),
|
| 63 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: checkpoint_interval=100000,
|
| 64 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: save_initial_state=False,
|
| 65 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: resume_checkpoint_path=None,
|
| 66 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: checkpoints_path_is_shared_file_system=False),
|
| 67 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: logging=LoggingArgs(log_level='info',
|
| 68 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: log_level_replica='info',
|
| 69 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: iteration_step_info_interval=1),
|
| 70 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: tokens=TokensArgs(sequence_length=4096,
|
| 71 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: train_steps=20,
|
| 72 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: micro_batch_size=16,
|
| 73 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: batch_accumulation_per_replica=32,
|
| 74 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: val_check_interval=-1,
|
| 75 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: limit_val_batches=0,
|
| 76 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: limit_test_batches=0),
|
| 77 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
|
| 78 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: adam_beta1=0.9,
|
| 79 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: adam_beta2=0.95,
|
| 80 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: torch_adam_is_fused=True,
|
| 81 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: name='adamW'),
|
| 82 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: zero_stage=1,
|
| 83 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: weight_decay=0.01,
|
| 84 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: clip_grad=1.0,
|
| 85 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: accumulate_grad_in_fp32=True,
|
| 86 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0001,
|
| 87 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: lr_warmup_steps=1,
|
| 88 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: lr_warmup_style='linear',
|
| 89 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: lr_decay_style='linear',
|
| 90 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: lr_decay_steps=19,
|
| 91 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: lr_decay_starting_step=None,
|
| 92 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: min_decay_lr=1e-05)),
|
| 93 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: data_stages=[DatasetStageArgs(name='Training Stage',
|
| 94 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: start_training_step=1,
|
| 95 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: data=DataArgs(dataset=PretrainDatasetsArgs(hf_dataset_or_datasets='roneneldan/TinyStories',
|
| 96 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: hf_dataset_splits='train',
|
| 97 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: hf_dataset_config_name=None,
|
| 98 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: dataset_processing_num_proc_per_process=64,
|
| 99 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: dataset_overwrite_cache=False,
|
| 100 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: text_column_name='text'),
|
| 101 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: seed=42,
|
| 102 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: num_loading_workers=0))],
|
| 103 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: profiler=ProfilerArgs(profiler_export_path=Path('/fsx/ferdinandmom/ferdinand-hf/bench_cluster/results/llama-1B/8_GPUS/dp-2_tp-1_pp-4_mbz-16')),
|
| 104 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: lighteval=None)
|
| 105 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: Model Config:
|
| 106 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: LlamaConfig(bos_token_id=1,
|
| 107 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: eos_token_id=2,
|
| 108 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: hidden_act='silu',
|
| 109 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: hidden_size=2048,
|
| 110 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: initializer_range=0.02,
|
| 111 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: intermediate_size=4096,
|
| 112 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: is_llama_config=True,
|
| 113 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: max_position_embeddings=4096,
|
| 114 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: num_attention_heads=32,
|
| 115 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: num_hidden_layers=24,
|
| 116 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: num_key_value_heads=32,
|
| 117 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: pad_token_id=None,
|
| 118 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: pretraining_tp=1,
|
| 119 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: rms_norm_eps=1e-05,
|
| 120 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: rope_scaling=None,
|
| 121 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: rope_theta=10000.0,
|
| 122 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: tie_word_embeddings=True,
|
| 123 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: use_cache=True,
|
| 124 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: vocab_size=50257)
|
| 125 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: Building model..
|
| 126 |
+
[default0]:07/03/2024 23:01:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: Setting PP block ranks...
|
| 127 |
+
[default3]:07/03/2024 23:01:58 [INFO|DP=1|PP=1|TP=0|ip-26-0-164-187]: No checkpoint path provided.
|
| 128 |
+
[default5]:07/03/2024 23:01:58 [INFO|DP=1|PP=2|TP=0|ip-26-0-164-187]: No checkpoint path provided.
|
| 129 |
+
[default1]:07/03/2024 23:01:58 [INFO|DP=1|PP=0|TP=0|ip-26-0-164-187]: No checkpoint path provided.
|
| 130 |
+
[default7]:07/03/2024 23:01:58 [INFO|DP=1|PP=3|TP=0|ip-26-0-164-187]: No checkpoint path provided.
|
| 131 |
+
[default2]:07/03/2024 23:01:59 [INFO|DP=0|PP=1|TP=0|ip-26-0-164-187]: Local number of parameters: 294M (560.05MiB)
|
| 132 |
+
[default2]:07/03/2024 23:01:59 [INFO|DP=0|PP=1|TP=0|ip-26-0-164-187]: [After model building] Memory usage: 567.07MiB. Peak allocated: 569.10MiB Peak reserved: 594.00MiB
|
| 133 |
+
[default2]:07/03/2024 23:01:59 [INFO|DP=0|PP=1|TP=0|ip-26-0-164-187]: No checkpoint path provided.
|
| 134 |
+
[default4]:07/03/2024 23:01:59 [INFO|DP=0|PP=2|TP=0|ip-26-0-164-187]: Local number of parameters: 252M (480.05MiB)
|
| 135 |
+
[default4]:07/03/2024 23:01:59 [INFO|DP=0|PP=2|TP=0|ip-26-0-164-187]: [After model building] Memory usage: 486.06MiB. Peak allocated: 488.09MiB Peak reserved: 502.00MiB
|
| 136 |
+
[default4]:07/03/2024 23:01:59 [INFO|DP=0|PP=2|TP=0|ip-26-0-164-187]: No checkpoint path provided.
|
| 137 |
+
[default6]:07/03/2024 23:01:59 [INFO|DP=0|PP=3|TP=0|ip-26-0-164-187]: Local number of parameters: 271M (516.35MiB)
|
| 138 |
+
[default6]:07/03/2024 23:01:59 [INFO|DP=0|PP=3|TP=0|ip-26-0-164-187]: [After model building] Memory usage: 520.36MiB. Peak allocated: 522.39MiB Peak reserved: 534.00MiB
|
| 139 |
+
[default6]:07/03/2024 23:01:59 [INFO|DP=0|PP=3|TP=0|ip-26-0-164-187]: No checkpoint path provided.
|
| 140 |
+
[default0]:07/03/2024 23:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: Total number of parameters: 1.21G (2312.82MiB)
|
| 141 |
+
[default0]:07/03/2024 23:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: Local number of parameters: 397M (756.37MiB)
|
| 142 |
+
[default0]:07/03/2024 23:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [After model building] Memory usage: 763.38MiB. Peak allocated: 765.41MiB Peak reserved: 792.00MiB
|
| 143 |
+
[default0]:07/03/2024 23:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: No checkpoint path provided.
|
| 144 |
+
[default0]:07/03/2024 23:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: Parametrizing model parameters using StandardParametrizator
|
| 145 |
+
[default0]:07/03/2024 23:02:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [Optimizer Building] Using LearningRateForSP as learning rate
|
| 146 |
+
[default0]:07/03/2024 23:02:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] Size of optimizer params per rank:
|
| 147 |
+
[default0]:07/03/2024 23:02:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 0 has 198M out of 397M (50.00%) params' optimizer states
|
| 148 |
+
[default0]:07/03/2024 23:02:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 1 has 198M out of 397M (50.00%) params' optimizer states
|
| 149 |
+
[default0]:07/03/2024 23:02:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [Training Plan] Stage Training Stage has 19 remaining training steps and has consumed 0 samples
|
| 150 |
+
[default0]:07/03/2024 23:02:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: Using `datasets` library
|
| 151 |
+
[default0]:07/03/2024 23:02:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: Loading tokenizer from openai-community/gpt2 and transformers/hf_hub versions ('4.41.2', '0.23.4')
|
| 152 |
+
[default0]:07/03/2024 23:02:03 [WARNING|DP=0|PP=0|TP=0|ip-26-0-164-187]: Repo card metadata block was not found. Setting CardData to empty.
|
| 153 |
+
[default0]:Repo card metadata block was not found. Setting CardData to empty.
|
| 154 |
+
[default0]:07/03/2024 23:02:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [Training Plan] There are 1 training stages
|
| 155 |
+
[default0]:07/03/2024 23:02:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [Stage Training Stage] start from step 1
|
| 156 |
+
[default0]:07/03/2024 23:02:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:
|
| 157 |
+
[default0]:07/03/2024 23:02:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [Start training] datetime: 2024-07-03 23:02:04.175922 | mbs: 16 | grad_accum: 32 | global_batch_size: 1024 | sequence_length: 4096 | train_steps: 20 | start_iteration_step: 0 | consumed_train_samples: 0
|
| 158 |
+
[default0]:07/03/2024 23:02:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: Resuming training from stage Training Stage, it has trained for 0 samples and has 19 remaining train steps
|
| 159 |
+
[default0]:07/03/2024 23:02:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: Memory usage: 3032.50MiB. Peak allocated 3032.50MiB. Peak reserved: 3064.00MiB
|
| 160 |
+
[default3]:07/03/2024 23:02:04 [WARNING|DP=1|PP=1|TP=0|ip-26-0-164-187]: Repo card metadata block was not found. Setting CardData to empty.
|
| 161 |
+
[default2]:07/03/2024 23:02:04 [WARNING|DP=0|PP=1|TP=0|ip-26-0-164-187]: Repo card metadata block was not found. Setting CardData to empty.
|
| 162 |
+
[default1]:07/03/2024 23:02:04 [WARNING|DP=1|PP=0|TP=0|ip-26-0-164-187]: Repo card metadata block was not found. Setting CardData to empty.
|
| 163 |
+
[default4]:07/03/2024 23:02:04 [WARNING|DP=0|PP=2|TP=0|ip-26-0-164-187]: Repo card metadata block was not found. Setting CardData to empty.
|
| 164 |
+
[default6]:07/03/2024 23:02:04 [WARNING|DP=0|PP=3|TP=0|ip-26-0-164-187]: Repo card metadata block was not found. Setting CardData to empty.
|
| 165 |
+
[default5]:07/03/2024 23:02:04 [WARNING|DP=1|PP=2|TP=0|ip-26-0-164-187]: Repo card metadata block was not found. Setting CardData to empty.
|
| 166 |
+
[default3]:Repo card metadata block was not found. Setting CardData to empty.
|
| 167 |
+
[default2]:Repo card metadata block was not found. Setting CardData to empty.
|
| 168 |
+
[default1]:Repo card metadata block was not found. Setting CardData to empty.
|
| 169 |
+
[default6]:Repo card metadata block was not found. Setting CardData to empty.
|
| 170 |
+
[default5]:Repo card metadata block was not found. Setting CardData to empty.
|
| 171 |
+
[default4]:Repo card metadata block was not found. Setting CardData to empty.
|
| 172 |
+
[default7]:07/03/2024 23:02:04 [WARNING|DP=1|PP=3|TP=0|ip-26-0-164-187]: Repo card metadata block was not found. Setting CardData to empty.
|
| 173 |
+
[default7]:Repo card metadata block was not found. Setting CardData to empty.
|
| 174 |
+
[default1]:[rank1]: Traceback (most recent call last):
|
| 175 |
+
[default1]:[rank1]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/run_train.py", line 237, in <module>
|
| 176 |
+
[default1]:[rank1]: trainer.train(dataloader)
|
| 177 |
+
[default1]:[rank1]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/trainer.py", line 429, in train
|
| 178 |
+
[default1]:[rank1]: outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
|
| 179 |
+
[default0]:[rank0]: Traceback (most recent call last):
|
| 180 |
+
[default0]:[rank0]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/run_train.py", line 237, in <module>
|
| 181 |
+
[default0]:[rank0]: trainer.train(dataloader)
|
| 182 |
+
[default0]:[rank0]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/trainer.py", line 429, in train
|
| 183 |
+
[default0]:[rank0]: outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
|
| 184 |
+
[default0]:[rank0]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/trainer.py", line 462, in training_step
|
| 185 |
+
[default0]:[rank0]: outputs = self.pipeline_engine.train_batch_iter(
|
| 186 |
+
[default0]:[rank0]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 252, in train_batch_iter
|
| 187 |
+
[default1]:[rank1]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/trainer.py", line 462, in training_step
|
| 188 |
+
[default1]:[rank1]: outputs = self.pipeline_engine.train_batch_iter(
|
| 189 |
+
[default0]:[rank0]: output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
|
| 190 |
+
[default0]:[rank0]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
|
| 191 |
+
[default0]:[rank0]: output = model(**micro_batch)
|
| 192 |
+
[default1]:[rank1]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 252, in train_batch_iter
|
| 193 |
+
[default0]:[rank0]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
|
| 194 |
+
[default0]:[rank0]: return self._call_impl(*args, **kwargs)
|
| 195 |
+
[default1]:[rank1]: output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
|
| 196 |
+
[default1]:[rank1]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
|
| 197 |
+
[default1]:[rank1]: output = model(**micro_batch)
|
| 198 |
+
[default1]:[rank1]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
|
| 199 |
+
[default1]:[rank1]: return self._call_impl(*args, **kwargs)
|
| 200 |
+
[default1]:[rank1]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
|
| 201 |
+
[default1]:[rank1]: return forward_call(*args, **kwargs)
|
| 202 |
+
[default1]:[rank1]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 891, in forward
|
| 203 |
+
[default1]:[rank1]: sharded_logits = self.model(
|
| 204 |
+
[default0]:[rank0]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
|
| 205 |
+
[default0]:[rank0]: return forward_call(*args, **kwargs)
|
| 206 |
+
[default0]:[rank0]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 891, in forward
|
| 207 |
+
[default0]:[rank0]: sharded_logits = self.model(
|
| 208 |
+
[default0]:[rank0]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
|
| 209 |
+
[default0]:[rank0]: return self._call_impl(*args, **kwargs)
|
| 210 |
+
[default0]:[rank0]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
|
| 211 |
+
[default0]:[rank0]: return forward_call(*args, **kwargs)
|
| 212 |
+
[default0]:[rank0]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 764, in forward
|
| 213 |
+
[default0]:[rank0]: return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
|
| 214 |
+
[default1]:[rank1]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
|
| 215 |
+
[default1]:[rank1]: return self._call_impl(*args, **kwargs)
|
| 216 |
+
[default1]:[rank1]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
|
| 217 |
+
[default1]:[rank1]: return forward_call(*args, **kwargs)
|
| 218 |
+
[default1]:[rank1]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 764, in forward
|
| 219 |
+
[default1]:[rank1]: return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
|
| 220 |
+
[default1]:[rank1]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 780, in forward_with_hidden_states
|
| 221 |
+
[default1]:[rank1]: hidden_encoder_states = encoder_block(**hidden_encoder_states)
|
| 222 |
+
[default1]:[rank1]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
|
| 223 |
+
[default0]:[rank0]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 780, in forward_with_hidden_states
|
| 224 |
+
[default0]:[rank0]: hidden_encoder_states = encoder_block(**hidden_encoder_states)
|
| 225 |
+
[default0]:[rank0]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
|
| 226 |
+
[default1]:[rank1]: return self._call_impl(*args, **kwargs)
|
| 227 |
+
[default0]:[rank0]: return self._call_impl(*args, **kwargs)
|
| 228 |
+
[default0]:[rank0]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
|
| 229 |
+
[default0]:[rank0]: return forward_call(*args, **kwargs)
|
| 230 |
+
[default1]:[rank1]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
|
| 231 |
+
[default0]:[rank0]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
|
| 232 |
+
[default0]:[rank0]: output = self.pp_block(**new_kwargs)
|
| 233 |
+
[default1]:[rank1]: return forward_call(*args, **kwargs)
|
| 234 |
+
[default0]:[rank0]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
|
| 235 |
+
[default0]:[rank0]: return self._call_impl(*args, **kwargs)
|
| 236 |
+
[default0]:[rank0]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
|
| 237 |
+
[default0]:[rank0]: return forward_call(*args, **kwargs)
|
| 238 |
+
[default1]:[rank1]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
|
| 239 |
+
[default1]:[rank1]: output = self.pp_block(**new_kwargs)
|
| 240 |
+
[default0]:[rank0]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 631, in forward
|
| 241 |
+
[default0]:[rank0]: output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
|
| 242 |
+
[default0]:[rank0]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
|
| 243 |
+
[default1]:[rank1]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
|
| 244 |
+
[default1]:[rank1]: return self._call_impl(*args, **kwargs)
|
| 245 |
+
[default1]:[rank1]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
|
| 246 |
+
[default1]:[rank1]: return forward_call(*args, **kwargs)
|
| 247 |
+
[default0]:[rank0]: return self._call_impl(*args, **kwargs)
|
| 248 |
+
[default1]:[rank1]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 631, in forward
|
| 249 |
+
[default0]:[rank0]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
|
| 250 |
+
[default1]:[rank1]: output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
|
| 251 |
+
[default0]:[rank0]: return forward_call(*args, **kwargs)
|
| 252 |
+
[default1]:[rank1]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
|
| 253 |
+
[default0]:[rank0]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 360, in forward
|
| 254 |
+
[default1]:[rank1]: return self._call_impl(*args, **kwargs)
|
| 255 |
+
[default0]:[rank0]: qkv_states = self.qkv_proj(
|
| 256 |
+
[default1]:[rank1]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
|
| 257 |
+
[default0]:[rank0]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
|
| 258 |
+
[default1]:[rank1]: return forward_call(*args, **kwargs)
|
| 259 |
+
[default1]:[rank1]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 360, in forward
|
| 260 |
+
[default0]:[rank0]: return self._call_impl(*args, **kwargs)
|
| 261 |
+
[default0]:[rank0]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
|
| 262 |
+
[default1]:[rank1]: qkv_states = self.qkv_proj(
|
| 263 |
+
[default1]:[rank1]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
|
| 264 |
+
[default1]:[rank1]: return self._call_impl(*args, **kwargs)
|
| 265 |
+
[default1]:[rank1]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
|
| 266 |
+
[default0]:[rank0]: return forward_call(*args, **kwargs)
|
| 267 |
+
[default1]:[rank1]: return forward_call(*args, **kwargs)
|
| 268 |
+
[default1]:[rank1]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 87, in forward
|
| 269 |
+
[default1]:[rank1]: return column_linear(
|
| 270 |
+
[default1]:[rank1]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 359, in column_linear
|
| 271 |
+
[default1]:[rank1]: return F.linear(input, weight, bias)
|
| 272 |
+
[default0]:[rank0]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 87, in forward
|
| 273 |
+
[default1]:[rank1]: torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 768.00 MiB. GPU has a total capacity of 79.33 GiB of which 623.94 MiB is free. Including non-PyTorch memory, this process has 78.71 GiB memory in use. Of the allocated memory 65.36 GiB is allocated by PyTorch, and 396.11 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation. See documentation for Memory Management (https://pytorch.org/docs/stable/notes/cuda.html#environment-variables)
|
| 274 |
+
[default0]:[rank0]: return column_linear(
|
| 275 |
+
[default0]:[rank0]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 359, in column_linear
|
| 276 |
+
[default0]:[rank0]: return F.linear(input, weight, bias)
|
| 277 |
+
[default0]:[rank0]: torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 768.00 MiB. GPU
|
| 278 |
+
[default6]:[rank6]: Traceback (most recent call last):
|
| 279 |
+
[default6]:[rank6]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/run_train.py", line 237, in <module>
|
| 280 |
+
[default6]:[rank6]: trainer.train(dataloader)
|
| 281 |
+
[default6]:[rank6]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/trainer.py", line 429, in train
|
| 282 |
+
[default6]:[rank6]: outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
|
| 283 |
+
[default6]:[rank6]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/trainer.py", line 462, in training_step
|
| 284 |
+
[default6]:[rank6]: outputs = self.pipeline_engine.train_batch_iter(
|
| 285 |
+
[default6]:[rank6]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
|
| 286 |
+
[default6]:[rank6]: output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
|
| 287 |
+
[default6]:[rank6]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
|
| 288 |
+
[default6]:[rank6]: output = model(**micro_batch)
|
| 289 |
+
[default6]:[rank6]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
|
| 290 |
+
[default6]:[rank6]: return self._call_impl(*args, **kwargs)
|
| 291 |
+
[default6]:[rank6]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
|
| 292 |
+
[default6]:[rank6]: return forward_call(*args, **kwargs)
|
| 293 |
+
[default6]:[rank6]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 891, in forward
|
| 294 |
+
[default6]:[rank6]: sharded_logits = self.model(
|
| 295 |
+
[default6]:[rank6]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
|
| 296 |
+
[default6]:[rank6]: return self._call_impl(*args, **kwargs)
|
| 297 |
+
[default6]:[rank6]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
|
| 298 |
+
[default6]:[rank6]: return forward_call(*args, **kwargs)
|
| 299 |
+
[default6]:[rank6]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 764, in forward
|
| 300 |
+
[default6]:[rank6]: return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
|
| 301 |
+
[default6]:[rank6]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 780, in forward_with_hidden_states
|
| 302 |
+
[default6]:[rank6]: hidden_encoder_states = encoder_block(**hidden_encoder_states)
|
| 303 |
+
[default6]:[rank6]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
|
| 304 |
+
[default6]:[rank6]: return self._call_impl(*args, **kwargs)
|
| 305 |
+
[default6]:[rank6]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
|
| 306 |
+
[default6]:[rank6]: return forward_call(*args, **kwargs)
|
| 307 |
+
[default6]:[rank6]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 126, in forward
|
| 308 |
+
[default6]:[rank6]: new_kwargs[name] = recv_from_pipeline_state_buffer(
|
| 309 |
+
[default6]:[rank6]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 117, in recv_from_pipeline_state_buffer
|
| 310 |
+
[default6]:[rank6]: pipeline_state.run_communication()
|
| 311 |
+
[default6]:[rank6]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/state.py", line 150, in run_communication
|
| 312 |
+
[default6]:[rank6]: recv_activation_tensor = recv_activation()
|
| 313 |
+
[default6]:[rank6]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/state.py", line 31, in __call__
|
| 314 |
+
[default6]:[rank6]: return self.p2p.recv_tensors(num_tensors=1, from_rank=self.from_rank)[0]
|
| 315 |
+
[default6]:[rank6]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 353, in recv_tensors
|
| 316 |
+
[default6]:[rank6]: buffers, futures = self.irecv_tensors(num_tensors=num_tensors, from_rank=from_rank, tag=tag)
|
| 317 |
+
[default6]:[rank6]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 326, in irecv_tensors
|
| 318 |
+
[default6]:[rank6]: meta = self._recv_meta(from_rank=from_rank, tag=tag)
|
| 319 |
+
[default6]:[rank6]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 246, in _recv_meta
|
| 320 |
+
[default6]:[rank6]: dist.recv(
|
| 321 |
+
[default6]:[rank6]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 75, in wrapper
|
| 322 |
+
[default6]:[rank6]: return func(*args, **kwargs)
|
| 323 |
+
[default6]:[rank6]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 1932, in recv
|
| 324 |
+
[default6]:[rank6]: pg.recv([tensor], group_src_rank, tag).wait()
|
| 325 |
+
[default6]:[rank6]: torch.distributed.DistBackendError: [3] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '2:3', but store->get('2:3') got error: Connection reset by peer
|
| 326 |
+
[default6]:[rank6]: Exception raised from recvBytes at ../torch/csrc/distributed/c10d/Utils.hpp:672 (most recent call first):
|
| 327 |
+
[default6]:[rank6]: frame #0: c10::Error::Error(c10::SourceLocation, std::string) + 0x57 (0x7f0f7b762897 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libc10.so)
|
| 328 |
+
[default6]:[rank6]: frame #1: <unknown function> + 0x5b3a23e (0x7f0fb527f23e in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 329 |
+
[default6]:[rank6]: frame #2: c10d::TCPStore::doWait(c10::ArrayRef<std::string>, std::chrono::duration<long, std::ratio<1l, 1000l> >) + 0x2c7 (0x7f0fb5279c87 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 330 |
+
[default6]:[rank6]: frame #3: c10d::TCPStore::doGet(std::string const&) + 0x32 (0x7f0fb5279f82 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 331 |
+
[default6]:[rank6]: frame #4: c10d::TCPStore::get(std::string const&) + 0xa1 (0x7f0fb527afd1 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 332 |
+
[default6]:[rank6]: frame #5: c10d::PrefixStore::get(std::string const&) + 0x31 (0x7f0fb522f371 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 333 |
+
[default6]:[rank6]: frame #6: c10d::PrefixStore::get(std::string const&) + 0x31 (0x7f0fb522f371 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 334 |
+
[default6]:[rank6]: frame #7: c10d::PrefixStore::get(std::string const&) + 0x31 (0x7f0fb522f371 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 335 |
+
[default6]:[rank6]: frame #8: c10d::PrefixStore::get(std::string const&) + 0x31 (0x7f0fb522f371 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 336 |
+
[default6]:[rank6]: frame #9: c10d::ProcessGroupNCCL::broadcastUniqueNCCLID(ncclUniqueId*, bool, std::string const&, int) + 0xa9 (0x7f0f7ca3c189 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
|
| 337 |
+
[default6]:[rank6]: frame #10: c10d::ProcessGroupNCCL::getNCCLComm(std::string const&, c10::Device&, c10d::OpType, int, bool) + 0xc50 (0x7f0f7ca43610 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
|
| 338 |
+
[default6]:[rank6]: frame #11: c10d::ProcessGroupNCCL::recv(std::vector<at::Tensor, std::allocator<at::Tensor> >&, int, int) + 0x5f8 (0x7f0f7ca62978 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
|
| 339 |
+
[default6]:[rank6]: frame #12: <unknown function> + 0x5adc309 (0x7f0fb5221309 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 340 |
+
[default6]:[rank6]: frame #13: <unknown function> + 0x5ae6f10 (0x7f0fb522bf10 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 341 |
+
[default6]:[rank6]: frame #14: <unknown function> + 0x5ae6fa5 (0x7f0fb522bfa5 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 342 |
+
[default6]:[rank6]: frame #15: <unknown function> + 0x5124446 (0x7f0fb4869446 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 343 |
+
[default6]:[rank6]: frame #16: <unknown function> + 0x1acf4b8 (0x7f0fb12144b8 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 344 |
+
[default6]:[rank6]: frame #17: <unknown function> + 0x5aee004 (0x7f0fb5233004 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 345 |
+
[default6]:[rank6]: frame #18: <unknown function> + 0x5af36b5 (0x7f0fb52386b5 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 346 |
+
[default6]:[rank6]: frame #19: <unknown function> + 0xd2631e (0x7f0fc7e2231e in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_python.so)
|
| 347 |
+
[default6]:[rank6]: frame #20: <unknown function> + 0x47def4 (0x7f0fc7579ef4 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_python.so)
|
| 348 |
+
[default6]:[rank6]: frame #21: <unknown function> + 0x1445a6 (0x55dd728355a6 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 349 |
+
[default6]:[rank6]: frame #22: _PyObject_MakeTpCall + 0x26b (0x55dd7282ea6b in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 350 |
+
[default6]:[rank6]: frame #23: <unknown function> + 0x150866 (0x55dd72841866 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 351 |
+
[default6]:[rank6]: frame #24: _PyEval_EvalFrameDefault + 0x4c12 (0x55dd7282a142 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 352 |
+
[default6]:[rank6]: frame #25: _PyFunction_Vectorcall + 0x6c (0x55dd72835a2c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 353 |
+
[default6]:[rank6]: frame #26: PyObject_Call + 0xbc (0x55dd72841f1c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 354 |
+
[default6]:[rank6]: frame #27: _PyEval_EvalFrameDefault + 0x2d83 (0x55dd728282b3 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 355 |
+
[default6]:[rank6]: frame #28: _PyFunction_Vectorcall + 0x6c (0x55dd72835a2c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 356 |
+
[default6]:[rank6]: frame #29: _PyEval_EvalFrameDefault + 0x13ca (0x55dd728268fa in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 357 |
+
[default6]:[rank6]: frame #30: <unknown function> + 0x150582 (0x55dd72841582 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 358 |
+
[default6]:[rank6]: frame #31: _PyEval_EvalFrameDefault + 0x13ca (0x55dd728268fa in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 359 |
+
[default6]:[rank6]: frame #32: <unknown function> + 0x150582 (0x55dd72841582 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 360 |
+
[default6]:[rank6]: frame #33: _PyEval_EvalFrameDefault + 0x13ca (0x55dd728268fa in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 361 |
+
[default6]:[rank6]: frame #34: <unknown function> + 0x150582 (0x55dd72841582 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 362 |
+
[default6]:[rank6]: frame #35: _PyEval_EvalFrameDefault + 0x13ca (0x55dd728268fa in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 363 |
+
[default6]:[rank6]: frame #36: _PyObject_FastCallDictTstate + 0xd0 (0x55dd7282df50 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 364 |
+
[default6]:[rank6]: frame #37: _PyObject_Call_Prepend + 0x69 (0x55dd7283fc39 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 365 |
+
[default6]:[rank6]: frame #38: <unknown function> + 0x211239 (0x55dd72902239 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 366 |
+
[default6]:[rank6]: frame #39: _PyObject_MakeTpCall + 0x26b (0x55dd7282ea6b in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 367 |
+
[default6]:[rank6]: frame #40: _PyEval_EvalFrameDefault + 0x4eb6 (0x55dd7282a3e6 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 368 |
+
[default6]:[rank6]: frame #41: _PyFunction_Vectorcall + 0x6c (0x55dd72835a2c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 369 |
+
[default6]:[rank6]: frame #42: _PyEval_EvalFrameDefault + 0x72c (0x55dd72825c5c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 370 |
+
[default6]:[rank6]: frame #43: _PyFunction_Vectorcall + 0x6c (0x55dd72835a2c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 371 |
+
[default6]:[rank6]: frame #44: _PyEval_EvalFrameDefault + 0x13ca (0x55dd728268fa in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 372 |
+
[default6]:[rank6]: frame #45: <unknown function> + 0x150582 (0x55dd72841582 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 373 |
+
[default6]:[rank6]: frame #46: PyObject_Call + 0xbc (0x55dd72841f1c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 374 |
+
[default6]:[rank6]: frame #47: _PyEval_EvalFrameDefault + 0x2d83 (0x55dd728282b3 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 375 |
+
[default6]:[rank6]: frame #48: <unknown function> + 0x150582 (0x55dd72841582 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 376 |
+
[default6]:[rank6]: frame #49: PyObject_Call + 0xbc (0x55dd72841f1c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 377 |
+
[default6]:[rank6]: frame #50: _PyEval_EvalFrameDefault + 0x2d83 (0x55dd728282b3 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 378 |
+
[default6]:[rank6]: frame #51: _PyFunction_Vectorcall + 0x6c (0x55dd72835a2c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 379 |
+
[default6]:[rank6]: frame #52: _PyObject_FastCallDictTstate + 0x187 (0x55dd7282e007 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 380 |
+
[default6]:[rank6]: frame #53: _PyObject_Call_Prepend + 0x69 (0x55dd7283fc39 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 381 |
+
[default6]:[rank6]: frame #54: <unknown function> + 0x211239 (0x55dd72902239 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 382 |
+
[default6]:[rank6]: frame #55: PyObject_Call + 0x207 (0x55dd72842067 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 383 |
+
[default6]:[rank6]: frame #56: _PyEval_EvalFrameDefault + 0x2d83 (0x55dd728282b3 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 384 |
+
[default6]:[rank6]: frame #57: <unknown function> + 0x150582 (0x55dd72841582 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 385 |
+
[default6]:[rank6]: frame #58: _PyEval_EvalFrameDefault + 0x13ca (0x55dd728268fa in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 386 |
+
[default6]:[rank6]: frame #59: <unknown function> + 0x150582 (0x55dd72841582 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 387 |
+
[default6]:[rank6]: frame #60: PyObject_Call + 0xbc (0x55dd72841f1c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 388 |
+
[default6]:[rank6]: frame #61: _PyEval_EvalFrameDefault + 0x2d83 (0x55dd728282b3 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 389 |
+
[default6]:[rank6]: frame #62: <unknown function> + 0x150582 (0x55dd72841582 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 390 |
+
[default6]:[rank6]: frame #63: PyObject_Call + 0xbc (0x55dd72841f1c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 391 |
+
[default6]:[rank6]: . This may indicate a possible application crash on rank 0 or a network set up issue.
|
| 392 |
+
[default7]:[rank7]: Traceback (most recent call last):
|
| 393 |
+
[default7]:[rank7]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/run_train.py", line 237, in <module>
|
| 394 |
+
[default7]:[rank7]: trainer.train(dataloader)
|
| 395 |
+
[default7]:[rank7]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/trainer.py", line 429, in train
|
| 396 |
+
[default7]:[rank7]: outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
|
| 397 |
+
[default7]:[rank7]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/trainer.py", line 462, in training_step
|
| 398 |
+
[default7]:[rank7]: outputs = self.pipeline_engine.train_batch_iter(
|
| 399 |
+
[default7]:[rank7]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
|
| 400 |
+
[default7]:[rank7]: output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
|
| 401 |
+
[default7]:[rank7]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
|
| 402 |
+
[default7]:[rank7]: output = model(**micro_batch)
|
| 403 |
+
[default7]:[rank7]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
|
| 404 |
+
[default7]:[rank7]: return self._call_impl(*args, **kwargs)
|
| 405 |
+
[default7]:[rank7]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
|
| 406 |
+
[default7]:[rank7]: return forward_call(*args, **kwargs)
|
| 407 |
+
[default7]:[rank7]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 891, in forward
|
| 408 |
+
[default7]:[rank7]: sharded_logits = self.model(
|
| 409 |
+
[default7]:[rank7]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
|
| 410 |
+
[default7]:[rank7]: return self._call_impl(*args, **kwargs)
|
| 411 |
+
[default7]:[rank7]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
|
| 412 |
+
[default7]:[rank7]: return forward_call(*args, **kwargs)
|
| 413 |
+
[default7]:[rank7]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 764, in forward
|
| 414 |
+
[default7]:[rank7]: return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
|
| 415 |
+
[default7]:[rank7]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 780, in forward_with_hidden_states
|
| 416 |
+
[default7]:[rank7]: hidden_encoder_states = encoder_block(**hidden_encoder_states)
|
| 417 |
+
[default7]:[rank7]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
|
| 418 |
+
[default7]:[rank7]: return self._call_impl(*args, **kwargs)
|
| 419 |
+
[default7]:[rank7]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
|
| 420 |
+
[default7]:[rank7]: return forward_call(*args, **kwargs)
|
| 421 |
+
[default7]:[rank7]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 126, in forward
|
| 422 |
+
[default7]:[rank7]: new_kwargs[name] = recv_from_pipeline_state_buffer(
|
| 423 |
+
[default7]:[rank7]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 117, in recv_from_pipeline_state_buffer
|
| 424 |
+
[default7]:[rank7]: pipeline_state.run_communication()
|
| 425 |
+
[default7]:[rank7]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/state.py", line 150, in run_communication
|
| 426 |
+
[default7]:[rank7]: recv_activation_tensor = recv_activation()
|
| 427 |
+
[default7]:[rank7]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/state.py", line 31, in __call__
|
| 428 |
+
[default7]:[rank7]: return self.p2p.recv_tensors(num_tensors=1, from_rank=self.from_rank)[0]
|
| 429 |
+
[default7]:[rank7]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 353, in recv_tensors
|
| 430 |
+
[default7]:[rank7]: buffers, futures = self.irecv_tensors(num_tensors=num_tensors, from_rank=from_rank, tag=tag)
|
| 431 |
+
[default7]:[rank7]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 326, in irecv_tensors
|
| 432 |
+
[default7]:[rank7]: meta = self._recv_meta(from_rank=from_rank, tag=tag)
|
| 433 |
+
[default7]:[rank7]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 246, in _recv_meta
|
| 434 |
+
[default7]:[rank7]: dist.recv(
|
| 435 |
+
[default7]:[rank7]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 75, in wrapper
|
| 436 |
+
[default7]:[rank7]: return func(*args, **kwargs)
|
| 437 |
+
[default7]:[rank7]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 1932, in recv
|
| 438 |
+
[default7]:[rank7]: pg.recv([tensor], group_src_rank, tag).wait()
|
| 439 |
+
[default7]:[rank7]: torch.distributed.DistBackendError: [3] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '2:3', but store->get('2:3') got error: Connection reset by peer
|
| 440 |
+
[default7]:[rank7]: Exception raised from recvBytes at ../torch/csrc/distributed/c10d/Utils.hpp:672 (most recent call first):
|
| 441 |
+
[default7]:[rank7]: frame #0: c10::Error::Error(c10::SourceLocation, std::string) + 0x57 (0x7fa8c59f8897 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libc10.so)
|
| 442 |
+
[default7]:[rank7]: frame #1: <unknown function> + 0x5b3a23e (0x7fa8ff51523e in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 443 |
+
[default7]:[rank7]: frame #2: c10d::TCPStore::doWait(c10::ArrayRef<std::string>, std::chrono::duration<long, std::ratio<1l, 1000l> >) + 0x2c7 (0x7fa8ff50fc87 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 444 |
+
[default7]:[rank7]: frame #3: c10d::TCPStore::doGet(std::string const&) + 0x32 (0x7fa8ff50ff82 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 445 |
+
[default7]:[rank7]: frame #4: c10d::TCPStore::get(std::string const&) + 0xa1 (0x7fa8ff510fd1 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 446 |
+
[default7]:[rank7]: frame #5: c10d::PrefixStore::get(std::string const&) + 0x31 (0x7fa8ff4c5371 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 447 |
+
[default7]:[rank7]: frame #6: c10d::PrefixStore::get(std::string const&) + 0x31 (0x7fa8ff4c5371 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 448 |
+
[default7]:[rank7]: frame #7: c10d::PrefixStore::get(std::string const&) + 0x31 (0x7fa8ff4c5371 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 449 |
+
[default7]:[rank7]: frame #8: c10d::PrefixStore::get(std::string const&) + 0x31 (0x7fa8ff4c5371 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 450 |
+
[default7]:[rank7]: frame #9: c10d::ProcessGroupNCCL::broadcastUniqueNCCLID(ncclUniqueId*, bool, std::string const&, int) + 0xa9 (0x7fa8c6cd2189 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
|
| 451 |
+
[default7]:[rank7]: frame #10: c10d::ProcessGroupNCCL::getNCCLComm(std::string const&, c10::Device&, c10d::OpType, int, bool) + 0xc50 (0x7fa8c6cd9610 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
|
| 452 |
+
[default7]:[rank7]: frame #11: c10d::ProcessGroupNCCL::recv(std::vector<at::Tensor, std::allocator<at::Tensor> >&, int, int) + 0x5f8 (0x7fa8c6cf8978 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
|
| 453 |
+
[default7]:[rank7]: frame #12: <unknown function> + 0x5adc309 (0x7fa8ff4b7309 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 454 |
+
[default7]:[rank7]: frame #13: <unknown function> + 0x5ae6f10 (0x7fa8ff4c1f10 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 455 |
+
[default7]:[rank7]: frame #14: <unknown function> + 0x5ae6fa5 (0x7fa8ff4c1fa5 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 456 |
+
[default7]:[rank7]: frame #15: <unknown function> + 0x5124446 (0x7fa8feaff446 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 457 |
+
[default7]:[rank7]: frame #16: <unknown function> + 0x1acf4b8 (0x7fa8fb4aa4b8 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 458 |
+
[default7]:[rank7]: frame #17: <unknown function> + 0x5aee004 (0x7fa8ff4c9004 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 459 |
+
[default7]:[rank7]: frame #18: <unknown function> + 0x5af36b5 (0x7fa8ff4ce6b5 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 460 |
+
[default7]:[rank7]: frame #19: <unknown function> + 0xd2631e (0x7fa9120b831e in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_python.so)
|
| 461 |
+
[default7]:[rank7]: frame #20: <unknown function> + 0x47def4 (0x7fa91180fef4 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_python.so)
|
| 462 |
+
[default7]:[rank7]: frame #21: <unknown function> + 0x1445a6 (0x55aaa0a195a6 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 463 |
+
[default7]:[rank7]: frame #22: _PyObject_MakeTpCall + 0x26b (0x55aaa0a12a6b in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 464 |
+
[default7]:[rank7]: frame #23: <unknown function> + 0x150866 (0x55aaa0a25866 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 465 |
+
[default7]:[rank7]: frame #24: _PyEval_EvalFrameDefault + 0x4c12 (0x55aaa0a0e142 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 466 |
+
[default7]:[rank7]: frame #25: _PyFunction_Vectorcall + 0x6c (0x55aaa0a19a2c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 467 |
+
[default7]:[rank7]: frame #26: PyObject_Call + 0xbc (0x55aaa0a25f1c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 468 |
+
[default7]:[rank7]: frame #27: _PyEval_EvalFrameDefault + 0x2d83 (0x55aaa0a0c2b3 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 469 |
+
[default7]:[rank7]: frame #28: _PyFunction_Vectorcall + 0x6c (0x55aaa0a19a2c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 470 |
+
[default7]:[rank7]: frame #29: _PyEval_EvalFrameDefault + 0x13ca (0x55aaa0a0a8fa in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 471 |
+
[default7]:[rank7]: frame #30: <unknown function> + 0x150582 (0x55aaa0a25582 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 472 |
+
[default7]:[rank7]: frame #31: _PyEval_EvalFrameDefault + 0x13ca (0x55aaa0a0a8fa in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 473 |
+
[default7]:[rank7]: frame #32: <unknown function> + 0x150582 (0x55aaa0a25582 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 474 |
+
[default7]:[rank7]: frame #33: _PyEval_EvalFrameDefault + 0x13ca (0x55aaa0a0a8fa in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 475 |
+
[default7]:[rank7]: frame #34: <unknown function> + 0x150582 (0x55aaa0a25582 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 476 |
+
[default7]:[rank7]: frame #35: _PyEval_EvalFrameDefault + 0x13ca (0x55aaa0a0a8fa in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 477 |
+
[default7]:[rank7]: frame #36: _PyObject_FastCallDictTstate + 0xd0 (0x55aaa0a11f50 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 478 |
+
[default7]:[rank7]: frame #37: _PyObject_Call_Prepend + 0x69 (0x55aaa0a23c39 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 479 |
+
[default7]:[rank7]: frame #38: <unknown function> + 0x211239 (0x55aaa0ae6239 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 480 |
+
[default7]:[rank7]: frame #39: _PyObject_MakeTpCall + 0x26b (0x55aaa0a12a6b in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 481 |
+
[default7]:[rank7]: frame #40: _PyEval_EvalFrameDefault + 0x4eb6 (0x55aaa0a0e3e6 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 482 |
+
[default7]:[rank7]: frame #41: _PyFunction_Vectorcall + 0x6c (0x55aaa0a19a2c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 483 |
+
[default7]:[rank7]: frame #42: _PyEval_EvalFrameDefault + 0x72c (0x55aaa0a09c5c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 484 |
+
[default7]:[rank7]: frame #43: _PyFunction_Vectorcall + 0x6c (0x55aaa0a19a2c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 485 |
+
[default7]:[rank7]: frame #44: _PyEval_EvalFrameDefault + 0x13ca (0x55aaa0a0a8fa in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 486 |
+
[default7]:[rank7]: frame #45: <unknown function> + 0x150582 (0x55aaa0a25582 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 487 |
+
[default7]:[rank7]: frame #46: PyObject_Call + 0xbc (0x55aaa0a25f1c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 488 |
+
[default7]:[rank7]: frame #47: _PyEval_EvalFrameDefault + 0x2d83 (0x55aaa0a0c2b3 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 489 |
+
[default7]:[rank7]: frame #48: <unknown function> + 0x150582 (0x55aaa0a25582 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 490 |
+
[default7]:[rank7]: frame #49: PyObject_Call + 0xbc (0x55aaa0a25f1c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 491 |
+
[default7]:[rank7]: frame #50: _PyEval_EvalFrameDefault + 0x2d83 (0x55aaa0a0c2b3 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 492 |
+
[default7]:[rank7]: frame #51: _PyFunction_Vectorcall + 0x6c (0x55aaa0a19a2c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 493 |
+
[default7]:[rank7]: frame #52: _PyObject_FastCallDictTstate + 0x187 (0x55aaa0a12007 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 494 |
+
[default7]:[rank7]: frame #53: _PyObject_Call_Prepend + 0x69 (0x55aaa0a23c39 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 495 |
+
[default7]:[rank7]: frame #54: <unknown function> + 0x211239 (0x55aaa0ae6239 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 496 |
+
[default7]:[rank7]: frame #55: PyObject_Call + 0x207 (0x55aaa0a26067 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 497 |
+
[default7]:[rank7]: frame #56: _PyEval_EvalFrameDefault + 0x2d83 (0x55aaa0a0c2b3 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 498 |
+
[default7]:[rank7]: frame #57: <unknown function> + 0x150582 (0x55aaa0a25582 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 499 |
+
[default7]:[rank7]: frame #58: _PyEval_EvalFrameDefault + 0x13ca (0x55aaa0a0a8fa in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 500 |
+
[default7]:[rank7]: frame #59: <unknown function> + 0x150582 (0x55aaa0a25582 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 501 |
+
[default7]:[rank7]: frame #60: PyObject_Call + 0xbc (0x55aaa0a25f1c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 502 |
+
[default7]:[rank7]: frame #61: _PyEval_EvalFrameDefault + 0x2d83 (0x55aaa0a0c2b3 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 503 |
+
[default7]:[rank7]: frame #62: <unknown function> + 0x150582 (0x55aaa0a25582 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 504 |
+
[default7]:[rank7]: frame #63: PyObject_Call + 0xbc (0x55aaa0a25f1c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 505 |
+
[default7]:[rank7]: . This may indicate a possible application crash on rank 0 or a network set up issue.
|
| 506 |
+
[default4]:[rank4]: Traceback (most recent call last):
|
| 507 |
+
[default4]:[rank4]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/run_train.py", line 237, in <module>
|
| 508 |
+
[default4]:[rank4]: trainer.train(dataloader)
|
| 509 |
+
[default4]:[rank4]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/trainer.py", line 429, in train
|
| 510 |
+
[default4]:[rank4]: outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
|
| 511 |
+
[default4]:[rank4]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/trainer.py", line 462, in training_step
|
| 512 |
+
[default4]:[rank4]: outputs = self.pipeline_engine.train_batch_iter(
|
| 513 |
+
[default4]:[rank4]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 252, in train_batch_iter
|
| 514 |
+
[default4]:[rank4]: output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
|
| 515 |
+
[default4]:[rank4]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
|
| 516 |
+
[default4]:[rank4]: output = model(**micro_batch)
|
| 517 |
+
[default4]:[rank4]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
|
| 518 |
+
[default4]:[rank4]: return self._call_impl(*args, **kwargs)
|
| 519 |
+
[default4]:[rank4]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
|
| 520 |
+
[default4]:[rank4]: return forward_call(*args, **kwargs)
|
| 521 |
+
[default4]:[rank4]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 891, in forward
|
| 522 |
+
[default4]:[rank4]: sharded_logits = self.model(
|
| 523 |
+
[default4]:[rank4]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
|
| 524 |
+
[default4]:[rank4]: return self._call_impl(*args, **kwargs)
|
| 525 |
+
[default4]:[rank4]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
|
| 526 |
+
[default4]:[rank4]: return forward_call(*args, **kwargs)
|
| 527 |
+
[default4]:[rank4]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 764, in forward
|
| 528 |
+
[default4]:[rank4]: return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
|
| 529 |
+
[default4]:[rank4]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/models/llama.py", line 780, in forward_with_hidden_states
|
| 530 |
+
[default4]:[rank4]: hidden_encoder_states = encoder_block(**hidden_encoder_states)
|
| 531 |
+
[default4]:[rank4]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
|
| 532 |
+
[default4]:[rank4]: return self._call_impl(*args, **kwargs)
|
| 533 |
+
[default4]:[rank4]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1541, in _call_impl
|
| 534 |
+
[default4]:[rank4]: return forward_call(*args, **kwargs)
|
| 535 |
+
[default4]:[rank4]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 126, in forward
|
| 536 |
+
[default4]:[rank4]: new_kwargs[name] = recv_from_pipeline_state_buffer(
|
| 537 |
+
[default4]:[rank4]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 117, in recv_from_pipeline_state_buffer
|
| 538 |
+
[default4]:[rank4]: pipeline_state.run_communication()
|
| 539 |
+
[default4]:[rank4]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/state.py", line 150, in run_communication
|
| 540 |
+
[default4]:[rank4]: recv_activation_tensor = recv_activation()
|
| 541 |
+
[default4]:[rank4]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/state.py", line 31, in __call__
|
| 542 |
+
[default4]:[rank4]: return self.p2p.recv_tensors(num_tensors=1, from_rank=self.from_rank)[0]
|
| 543 |
+
[default4]:[rank4]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 353, in recv_tensors
|
| 544 |
+
[default4]:[rank4]: buffers, futures = self.irecv_tensors(num_tensors=num_tensors, from_rank=from_rank, tag=tag)
|
| 545 |
+
[default4]:[rank4]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 326, in irecv_tensors
|
| 546 |
+
[default4]:[rank4]: meta = self._recv_meta(from_rank=from_rank, tag=tag)
|
| 547 |
+
[default4]:[rank4]: File "/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 246, in _recv_meta
|
| 548 |
+
[default4]:[rank4]: dist.recv(
|
| 549 |
+
[default4]:[rank4]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 75, in wrapper
|
| 550 |
+
[default4]:[rank4]: return func(*args, **kwargs)
|
| 551 |
+
[default4]:[rank4]: File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 1932, in recv
|
| 552 |
+
[default4]:[rank4]: pg.recv([tensor], group_src_rank, tag).wait()
|
| 553 |
+
[default4]:[rank4]: torch.distributed.DistBackendError: [2] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '1:2', but store->get('1:2') got error: Connection reset by peer
|
| 554 |
+
[default4]:[rank4]: Exception raised from recvBytes at ../torch/csrc/distributed/c10d/Utils.hpp:672 (most recent call first):
|
| 555 |
+
[default4]:[rank4]: frame #0: c10::Error::Error(c10::SourceLocation, std::string) + 0x57 (0x7f00a79c7897 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libc10.so)
|
| 556 |
+
[default4]:[rank4]: frame #1: <unknown function> + 0x5b3a23e (0x7f00e14e423e in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 557 |
+
[default4]:[rank4]: frame #2: c10d::TCPStore::doWait(c10::ArrayRef<std::string>, std::chrono::duration<long, std::ratio<1l, 1000l> >) + 0x2c7 (0x7f00e14dec87 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 558 |
+
[default4]:[rank4]: frame #3: c10d::TCPStore::doGet(std::string const&) + 0x32 (0x7f00e14def82 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 559 |
+
[default4]:[rank4]: frame #4: c10d::TCPStore::get(std::string const&) + 0xa1 (0x7f00e14dffd1 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 560 |
+
[default4]:[rank4]: frame #5: c10d::PrefixStore::get(std::string const&) + 0x31 (0x7f00e1494371 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 561 |
+
[default4]:[rank4]: frame #6: c10d::PrefixStore::get(std::string const&) + 0x31 (0x7f00e1494371 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 562 |
+
[default4]:[rank4]: frame #7: c10d::PrefixStore::get(std::string const&) + 0x31 (0x7f00e1494371 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 563 |
+
[default4]:[rank4]: frame #8: c10d::PrefixStore::get(std::string const&) + 0x31 (0x7f00e1494371 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 564 |
+
[default4]:[rank4]: frame #9: c10d::ProcessGroupNCCL::broadcastUniqueNCCLID(ncclUniqueId*, bool, std::string const&, int) + 0xa9 (0x7f00a8ca1189 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
|
| 565 |
+
[default4]:[rank4]: frame #10: c10d::ProcessGroupNCCL::getNCCLComm(std::string const&, c10::Device&, c10d::OpType, int, bool) + 0xc50 (0x7f00a8ca8610 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
|
| 566 |
+
[default4]:[rank4]: frame #11: c10d::ProcessGroupNCCL::recv(std::vector<at::Tensor, std::allocator<at::Tensor> >&, int, int) + 0x5f8 (0x7f00a8cc7978 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
|
| 567 |
+
[default4]:[rank4]: frame #12: <unknown function> + 0x5adc309 (0x7f00e1486309 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 568 |
+
[default4]:[rank4]: frame #13: <unknown function> + 0x5ae6f10 (0x7f00e1490f10 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 569 |
+
[default4]:[rank4]: frame #14: <unknown function> + 0x5ae6fa5 (0x7f00e1490fa5 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 570 |
+
[default4]:[rank4]: frame #15: <unknown function> + 0x5124446 (0x7f00e0ace446 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 571 |
+
[default4]:[rank4]: frame #16: <unknown function> + 0x1acf4b8 (0x7f00dd4794b8 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 572 |
+
[default4]:[rank4]: frame #17: <unknown function> + 0x5aee004 (0x7f00e1498004 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 573 |
+
[default4]:[rank4]: frame #18: <unknown function> + 0x5af36b5 (0x7f00e149d6b5 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_cpu.so)
|
| 574 |
+
[default4]:[rank4]: frame #19: <unknown function> + 0xd2631e (0x7f00f408731e in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_python.so)
|
| 575 |
+
[default4]:[rank4]: frame #20: <unknown function> + 0x47def4 (0x7f00f37deef4 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/lib/libtorch_python.so)
|
| 576 |
+
[default4]:[rank4]: frame #21: <unknown function> + 0x1445a6 (0x562108c775a6 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 577 |
+
[default4]:[rank4]: frame #22: _PyObject_MakeTpCall + 0x26b (0x562108c70a6b in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 578 |
+
[default4]:[rank4]: frame #23: <unknown function> + 0x150866 (0x562108c83866 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 579 |
+
[default4]:[rank4]: frame #24: _PyEval_EvalFrameDefault + 0x4c12 (0x562108c6c142 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 580 |
+
[default4]:[rank4]: frame #25: _PyFunction_Vectorcall + 0x6c (0x562108c77a2c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 581 |
+
[default4]:[rank4]: frame #26: PyObject_Call + 0xbc (0x562108c83f1c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 582 |
+
[default4]:[rank4]: frame #27: _PyEval_EvalFrameDefault + 0x2d83 (0x562108c6a2b3 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 583 |
+
[default4]:[rank4]: frame #28: _PyFunction_Vectorcall + 0x6c (0x562108c77a2c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 584 |
+
[default4]:[rank4]: frame #29: _PyEval_EvalFrameDefault + 0x13ca (0x562108c688fa in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 585 |
+
[default4]:[rank4]: frame #30: <unknown function> + 0x150582 (0x562108c83582 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 586 |
+
[default4]:[rank4]: frame #31: _PyEval_EvalFrameDefault + 0x13ca (0x562108c688fa in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 587 |
+
[default4]:[rank4]: frame #32: <unknown function> + 0x150582 (0x562108c83582 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 588 |
+
[default4]:[rank4]: frame #33: _PyEval_EvalFrameDefault + 0x13ca (0x562108c688fa in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 589 |
+
[default4]:[rank4]: frame #34: <unknown function> + 0x150582 (0x562108c83582 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 590 |
+
[default4]:[rank4]: frame #35: _PyEval_EvalFrameDefault + 0x13ca (0x562108c688fa in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 591 |
+
[default4]:[rank4]: frame #36: _PyObject_FastCallDictTstate + 0xd0 (0x562108c6ff50 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 592 |
+
[default4]:[rank4]: frame #37: _PyObject_Call_Prepend + 0x69 (0x562108c81c39 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 593 |
+
[default4]:[rank4]: frame #38: <unknown function> + 0x211239 (0x562108d44239 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 594 |
+
[default4]:[rank4]: frame #39: _PyObject_MakeTpCall + 0x26b (0x562108c70a6b in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 595 |
+
[default4]:[rank4]: frame #40: _PyEval_EvalFrameDefault + 0x4eb6 (0x562108c6c3e6 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 596 |
+
[default4]:[rank4]: frame #41: _PyFunction_Vectorcall + 0x6c (0x562108c77a2c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 597 |
+
[default4]:[rank4]: frame #42: _PyEval_EvalFrameDefault + 0x72c (0x562108c67c5c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 598 |
+
[default4]:[rank4]: frame #43: _PyFunction_Vectorcall + 0x6c (0x562108c77a2c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 599 |
+
[default4]:[rank4]: frame #44: _PyEval_EvalFrameDefault + 0x13ca (0x562108c688fa in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 600 |
+
[default4]:[rank4]: frame #45: <unknown function> + 0x150582 (0x562108c83582 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 601 |
+
[default4]:[rank4]: frame #46: PyObject_Call + 0xbc (0x562108c83f1c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 602 |
+
[default4]:[rank4]: frame #47: _PyEval_EvalFrameDefault + 0x2d83 (0x562108c6a2b3 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 603 |
+
[default4]:[rank4]: frame #48: <unknown function> + 0x150582 (0x562108c83582 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 604 |
+
[default4]:[rank4]: frame #49: PyObject_Call + 0xbc (0x562108c83f1c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 605 |
+
[default4]:[rank4]: frame #50: _PyEval_EvalFrameDefault + 0x2d83 (0x562108c6a2b3 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 606 |
+
[default4]:[rank4]: frame #51: _PyFunction_Vectorcall + 0x6c (0x562108c77a2c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 607 |
+
[default4]:[rank4]: frame #52: _PyObject_FastCallDictTstate + 0x187 (0x562108c70007 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 608 |
+
[default4]:[rank4]: frame #53: _PyObject_Call_Prepend + 0x69 (0x562108c81c39 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 609 |
+
[default4]:[rank4]: frame #54: <unknown function> + 0x211239 (0x562108d44239 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 610 |
+
[default4]:[rank4]: frame #55: PyObject_Call + 0x207 (0x562108c84067 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 611 |
+
[default4]:[rank4]: frame #56: _PyEval_EvalFrameDefault + 0x2d83 (0x562108c6a2b3 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 612 |
+
[default4]:[rank4]: frame #57: <unknown function> + 0x150582 (0x562108c83582 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 613 |
+
[default4]:[rank4]: frame #58: _PyEval_EvalFrameDefault + 0x13ca (0x562108c688fa in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 614 |
+
[default4]:[rank4]: frame #59: <unknown function> + 0x150582 (0x562108c83582 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 615 |
+
[default4]:[rank4]: frame #60: PyObject_Call + 0xbc (0x562108c83f1c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 616 |
+
[default4]:[rank4]: frame #61: _PyEval_EvalFrameDefault + 0x2d83 (0x562108c6a2b3 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 617 |
+
[default4]:[rank4]: frame #62: <unknown function> + 0x150582 (0x562108c83582 in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 618 |
+
[default4]:[rank4]: frame #63: PyObject_Call + 0xbc (0x562108c83f1c in /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10)
|
| 619 |
+
[default4]:[rank4]: . This may indicate a possible application crash on rank 0 or a network set up issue.
|
| 620 |
+
W0703 23:02:15.311000 139658830735168 torch/distributed/elastic/multiprocessing/api.py:851] Sending process 24320 closing signal SIGTERM
|
| 621 |
+
W0703 23:02:15.311000 139658830735168 torch/distributed/elastic/multiprocessing/api.py:851] Sending process 24322 closing signal SIGTERM
|
| 622 |
+
W0703 23:02:15.311000 139658830735168 torch/distributed/elastic/multiprocessing/api.py:851] Sending process 24324 closing signal SIGTERM
|
| 623 |
+
W0703 23:02:15.312000 139658830735168 torch/distributed/elastic/multiprocessing/api.py:851] Sending process 24325 closing signal SIGTERM
|
| 624 |
+
W0703 23:02:15.313000 139658830735168 torch/distributed/elastic/multiprocessing/api.py:851] Sending process 24327 closing signal SIGTERM
|
| 625 |
+
W0703 23:02:15.314000 139658830735168 torch/distributed/elastic/multiprocessing/api.py:851] Sending process 24329 closing signal SIGTERM
|
| 626 |
+
E0703 23:02:17.031000 139658830735168 torch/distributed/elastic/multiprocessing/api.py:826] failed (exitcode: 1) local_rank: 0 (pid: 24317) of binary: /fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/python3.10
|
| 627 |
+
Traceback (most recent call last):
|
| 628 |
+
File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/bin/torchrun", line 8, in <module>
|
| 629 |
+
sys.exit(main())
|
| 630 |
+
File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 347, in wrapper
|
| 631 |
+
return f(*args, **kwargs)
|
| 632 |
+
File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/distributed/run.py", line 879, in main
|
| 633 |
+
run(args)
|
| 634 |
+
File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/distributed/run.py", line 870, in run
|
| 635 |
+
elastic_launch(
|
| 636 |
+
File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 132, in __call__
|
| 637 |
+
return launch_agent(self._config, self._entrypoint, list(args))
|
| 638 |
+
File "/fsx/ferdinandmom/miniforge3/envs/env-bench-cluster/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 263, in launch_agent
|
| 639 |
+
raise ChildFailedError(
|
| 640 |
+
torch.distributed.elastic.multiprocessing.errors.ChildFailedError:
|
| 641 |
+
============================================================
|
| 642 |
+
/fsx/ferdinandmom/ferdinand-hf/bench_cluster/nanotron/run_train.py FAILED
|
| 643 |
+
------------------------------------------------------------
|
| 644 |
+
Failures:
|
| 645 |
+
[1]:
|
| 646 |
+
time : 2024-07-03_23:02:15
|
| 647 |
+
host : ip-26-0-164-187.ec2.internal
|
| 648 |
+
rank : 1 (local_rank: 1)
|
| 649 |
+
exitcode : 1 (pid: 24319)
|
| 650 |
+
error_file: <N/A>
|
| 651 |
+
traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
|
| 652 |
+
------------------------------------------------------------
|
| 653 |
+
Root Cause (first observed failure):
|
| 654 |
+
[0]:
|
| 655 |
+
time : 2024-07-03_23:02:15
|
| 656 |
+
host : ip-26-0-164-187.ec2.internal
|
| 657 |
+
rank : 0 (local_rank: 0)
|
| 658 |
+
exitcode : 1 (pid: 24317)
|
| 659 |
+
error_file: <N/A>
|
| 660 |
+
traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
|
| 661 |
+
============================================================
|
| 662 |
+
srun: error: ip-26-0-164-187: task 0: Exited with exit code 1
|
| 663 |
+
Consider using `hf_transfer` for faster uploads. This solution comes with some limitations. See https://huggingface.co/docs/huggingface_hub/hf_transfer for more details.
|
llama-1B/8_GPUS/dp-2_tp-1_pp-4_mbz-16/status.txt
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
oom
|