Create config.json
Browse files- config.json +38 -0
config.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model_type": "halo_s",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"HaloSForCausalLM"
|
| 5 |
+
],
|
| 6 |
+
"auto_map": {
|
| 7 |
+
"AutoConfig": "configuration_halos.HaloSConfig",
|
| 8 |
+
"AutoModel": "modeling_halos.HaloSHFModel",
|
| 9 |
+
"AutoModelForCausalLM": "modeling_halos.HaloSForCausalLM"
|
| 10 |
+
},
|
| 11 |
+
"vocab_size": 50257,
|
| 12 |
+
"hidden_size": 512,
|
| 13 |
+
"num_layers": 6,
|
| 14 |
+
"num_heads": 8,
|
| 15 |
+
"num_kv_heads": 2,
|
| 16 |
+
"num_globals": 2,
|
| 17 |
+
"local_window": 64,
|
| 18 |
+
"dilated_offsets": [1, 2, 4, 8, 16, 32, 64, 128],
|
| 19 |
+
"num_random": 2,
|
| 20 |
+
"dropout": 0.1,
|
| 21 |
+
"use_swiglu": false,
|
| 22 |
+
"max_seq_len": 1024,
|
| 23 |
+
"torch_dtype": "float16",
|
| 24 |
+
"framework": "pyhalos",
|
| 25 |
+
"architecture": "HALO-S",
|
| 26 |
+
"tokenizer": "tiktoken-gpt2",
|
| 27 |
+
"training": {
|
| 28 |
+
"dataset": "wikitext-103-raw-v1",
|
| 29 |
+
"max_tokens": 30000000,
|
| 30 |
+
"epochs": 2,
|
| 31 |
+
"batch_size": 4,
|
| 32 |
+
"grad_accumulation": 4,
|
| 33 |
+
"effective_batch": 32,
|
| 34 |
+
"precision": "fp16_mixed",
|
| 35 |
+
"gpus": "2x T4 (DataParallel)",
|
| 36 |
+
"gradient_checkpointing": true
|
| 37 |
+
}
|
| 38 |
+
}
|