Kastan's group workspace
Group: Oct-04__16:06
Name
1 visualized
State
Notes
User
Tags
Created
Runtime
Sweep
BATCH_SIZE
LEARNING_RATE
LOG_PATH
NUM_EPOCHS
SEQ_LENGTH
TOTAL_BATCH_SIZE
VOCAB_SIZE
WARMUP_EPOCHS
WEIGHT_DECAY
clip_grad_norm
conda_env_name
data_dir
fp16.mode
gpt2_8B
gpt2_large
gpt2_medium
gpt2_xl
gradient_accumulation
model.checkpoint
model.decoder_dtype
model.embed_dtype
model.head_dtype
model.layernorm_dtype
model.max_position_embeddings
model.vocab_size
model_dtypes.decoder_dtype
model_dtypes.embed_dtype
model_dtypes.head_dtype
model_dtypes.layernorm_dtype
num_gpus_per_node
optimizer.lr
optimizer.weight_decay
parallel.pipeline
quant_gpt2_8B
quant_gpt2_micro
quant_gpt2_small
quant_gpt2_xl
total_gpus
MICRO_BATCH_SIZE
NUM_MICRO_BATCHES
PIPELINE_SIZE
TENSOR_PARALLEL_MODE
TENSOR_PARALLEL_SIZE
loss.type
Finished
-
kastan
BATCH_SIZE4
NUM_EPOCHS=2
Oct-04__16:06
SLURM=958216
WORLD_SIZE=1
7m 1s
-
4
0.00015
./gpt2_bs4_lr0.00015/
2
256
4
50304
1
0.01
-
col_ai_quant
/u/kastanday/LLM-Distributed-Quantization/datasets/small-gpt-dataset.json
-
-
-
-
-
-
false
torch.bfloat16
torch.bfloat16
torch.bfloat16
torch.bfloat16
256
50304
torch.bfloat16
torch.bfloat16
torch.bfloat16
torch.bfloat16
1
0.00015
0.01
1
-
-
titans.model.quant_gpt.quant_gpt.quant_gpt2_small
-
1
-
-
-
-
-
-
1-1
of 1