initial 70M experiment

configs/scaling-0.1/70M_512step.yml

@@ -0,0 +1,101 @@
+{
+  "pipe_parallel_size": 1,
+  "model_parallel_size": 1,
+
+  "num_layers": 6,
+  "hidden_size": 512,
+  "num_attention_heads": 8,
+  "seq_length": 2048,
+  "max_position_embeddings": 2048,
+  "pos_emb": "rotary",
+  "rotary_pct": 0.25,
+  "no_weight_tying": true,
+  "gpt_j_residual": true,
+  "output_layer_parallelism": "column",
+
+  "attention_config": [[["flash"], 6]],
+
+  "scaled_upper_triang_masked_softmax_fusion": true,
+  "bias_gelu_fusion": true,
+
+  "init_method": "small_init",
+  "output_layer_init_method": "wang_init",
+
+  "optimizer": {
+    "type": "Adam",
+    "params": {
+      "lr": 0.001,
+      "betas": [0.9, 0.95],
+      "eps": 1.0e-8
+    }
+  },
+  "min_lr": 0.0001,
+
+  "zero_optimization": {
+    "stage": 1,
+    "allgather_partitions": true,
+    "allgather_bucket_size": 500000000,
+    "overlap_comm": true,
+    "reduce_scatter": true,
+    "reduce_bucket_size": 500000000,
+    "contiguous_gradients": true,
+    "cpu_offload": false
+  },
+
+  "train_micro_batch_size_per_gpu": 64,
+  "gas": 2,
+  "data_impl": "mmap",
+  "num_workers": 1,
+
+  "checkpoint_activations": true,
+  "checkpoint_num_layers": 1,
+  "partition_activations": true,
+  "synchronize_each_layer": true,
+
+  "gradient_clipping": 1.0,
+  "weight_decay": 0.1,
+  "hidden_dropout": 0,
+  "attention_dropout": 0,
+
+  "fp16": {
+    "fp16": true,
+    "enabled": true,
+    "loss_scale": 0,
+    "loss_scale_window": 1000,
+    "initial_scale_power": 12,
+    "hysteresis": 2,
+    "min_loss_scale": 1
+  },
+
+   "distributed_backend": "nccl",
+
+   "train_iters": 512,
+   "lr_decay_iters": 512,
+   "lr_decay_style": "cosine",
+   "warmup_iter": 100,
+   "checkpoint_factor": 512,
+   "extra_save_iters": [1],
+   "eval_interval": 128,
+   "eval_iters": 50,
+
+   "log_interval": 1,
+   "steps_per_print": 1,
+   "wall_clock_breakdown": true,
+
+   "tokenizer_type": "HFTokenizer",
+   "vocab-file": "/home/za2514/compute/pythia_tokenizer/tokenizer.json",
+
+   "save": "/home/za2514/compute/saved-weights/scaling-0.1/",
+
+   "log-dir": "/home/za2514/compute/scaling/gpt-neox/logs/scaling-0.1",
+
+   "use_wandb": true,
+   "wandb_group": "70M-512step",
+   "wandb_project": "scaling-0.1",
+   "wandb_team": "math-lm",
+   "wandb_host": "https://api.wandb.ai",
+
+   "train-data-paths": ["/home/za2514/compute/data/proof-pile_llama/train/arxiv-rp/arxiv-rp_text_document"],
+   "valid-data-paths": ["/home/za2514/compute/data/proof-pile_llama/validation/arxiv-rp/arxiv-rp_text_document"],
+   "test-data-paths": ["/home/za2514/compute/data/proof-pile_llama/test/arxiv-rp/arxiv-rp_text_document"]
+}