Upload folder using huggingface_hub
Browse files
step_000000004000/config.json
ADDED
|
@@ -0,0 +1,37 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"ItttModel"
|
| 4 |
+
],
|
| 5 |
+
"attention_bias": false,
|
| 6 |
+
"attention_dropout": 0.0,
|
| 7 |
+
"base_lr": 0.001,
|
| 8 |
+
"base_model": "HuggingFaceTB/SmolLM2-360M",
|
| 9 |
+
"bos_token_id": 0,
|
| 10 |
+
"disable_fast_weights": false,
|
| 11 |
+
"dtype": "float32",
|
| 12 |
+
"eos_token_id": 0,
|
| 13 |
+
"fast_weight_size": 960,
|
| 14 |
+
"head_dim": 64,
|
| 15 |
+
"hidden_act": "silu",
|
| 16 |
+
"hidden_size": 960,
|
| 17 |
+
"initializer_range": 0.02,
|
| 18 |
+
"intermediate_size": 2560,
|
| 19 |
+
"is_llama_config": true,
|
| 20 |
+
"max_position_embeddings": 8192,
|
| 21 |
+
"mlp_bias": false,
|
| 22 |
+
"model_type": "ittt",
|
| 23 |
+
"momentum_beta": 0.9,
|
| 24 |
+
"momentum_dtype": "bfloat16",
|
| 25 |
+
"num_attention_heads": 15,
|
| 26 |
+
"num_hidden_layers": 32,
|
| 27 |
+
"num_key_value_heads": 5,
|
| 28 |
+
"pretraining_tp": 1,
|
| 29 |
+
"rms_norm_eps": 1e-05,
|
| 30 |
+
"rope_interleaved": false,
|
| 31 |
+
"rope_scaling": null,
|
| 32 |
+
"rope_theta": 100000,
|
| 33 |
+
"state_dtype": "float32",
|
| 34 |
+
"transformers_version": "4.57.3",
|
| 35 |
+
"use_cache": true,
|
| 36 |
+
"vocab_size": 49152
|
| 37 |
+
}
|
step_000000004000/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7c9099fa03e96ab0a41a4e58caaf6da0beec40a36c8c16bb3764b924bd3a3a41
|
| 3 |
+
size 2155129456
|