Upload folder using huggingface_hub
Browse files- 000000020000/config.json +102 -0
- 000000020000/model.pt +3 -0
000000020000/config.json
ADDED
|
@@ -0,0 +1,102 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"pure_modules": [],
|
| 3 |
+
"remat": {
|
| 4 |
+
"activation_checkpoint_layers": [
|
| 5 |
+
"LlamaDecoderLayer"
|
| 6 |
+
],
|
| 7 |
+
"scan_layers": "model.layers",
|
| 8 |
+
"offload_tensors": [
|
| 9 |
+
"decoder_input"
|
| 10 |
+
],
|
| 11 |
+
"optimization_barrier_layers": [
|
| 12 |
+
"LlamaDecoderLayer"
|
| 13 |
+
]
|
| 14 |
+
},
|
| 15 |
+
"type": "custom_llama.CustomLlamaForCausalLM",
|
| 16 |
+
"pretrained_url": "aklein4/SmolLM2-1.7B-TPU",
|
| 17 |
+
"pretrained_step": 0,
|
| 18 |
+
"pretrained_strict": false,
|
| 19 |
+
"torch_dtype": "float32",
|
| 20 |
+
"vocab_size": 49152,
|
| 21 |
+
"bos_token_id": 0,
|
| 22 |
+
"eos_token_id": 0,
|
| 23 |
+
"pad_token_id": 49152,
|
| 24 |
+
"hidden_size": 2048,
|
| 25 |
+
"num_hidden_layers": 24,
|
| 26 |
+
"num_attention_heads": 32,
|
| 27 |
+
"num_key_value_heads": 32,
|
| 28 |
+
"intermediate_size": 8192,
|
| 29 |
+
"hidden_act": "silu",
|
| 30 |
+
"max_position_embeddings": 8192,
|
| 31 |
+
"rope_theta": 130000,
|
| 32 |
+
"initializer_range": null,
|
| 33 |
+
"gaussian_init": true,
|
| 34 |
+
"attention_dropout": false,
|
| 35 |
+
"attention_bias": false,
|
| 36 |
+
"rms_norm_eps": 1e-05,
|
| 37 |
+
"pad_attention_bias_value": -100.0,
|
| 38 |
+
"attention_kernel": "flash_attention",
|
| 39 |
+
"sharding": {
|
| 40 |
+
"model.embed_tokens.weight": [
|
| 41 |
+
"fsdp",
|
| 42 |
+
null
|
| 43 |
+
],
|
| 44 |
+
"lm_head.weight": [
|
| 45 |
+
"fsdp",
|
| 46 |
+
null
|
| 47 |
+
],
|
| 48 |
+
"model.layers.*.self_attn.q_proj.weight": [
|
| 49 |
+
"fsdp",
|
| 50 |
+
null
|
| 51 |
+
],
|
| 52 |
+
"model.layers.*.self_attn.k_proj.weight": [
|
| 53 |
+
"fsdp",
|
| 54 |
+
null
|
| 55 |
+
],
|
| 56 |
+
"model.layers.*.self_attn.v_proj.weight": [
|
| 57 |
+
"fsdp",
|
| 58 |
+
null
|
| 59 |
+
],
|
| 60 |
+
"model.layers.*.self_attn.o_proj.weight": [
|
| 61 |
+
"fsdp",
|
| 62 |
+
null
|
| 63 |
+
],
|
| 64 |
+
"model.layers.*.mlp.gate_proj.weight": [
|
| 65 |
+
"fsdp",
|
| 66 |
+
null
|
| 67 |
+
],
|
| 68 |
+
"model.layers.*.mlp.up_proj.weight": [
|
| 69 |
+
"fsdp",
|
| 70 |
+
null
|
| 71 |
+
],
|
| 72 |
+
"model.layers.*.mlp.down_proj.weight": [
|
| 73 |
+
null,
|
| 74 |
+
"fsdp"
|
| 75 |
+
],
|
| 76 |
+
"model.layers.*.input_layernorm.weight": [
|
| 77 |
+
"fsdp"
|
| 78 |
+
],
|
| 79 |
+
"model.layers.*.post_attention_layernorm.weight": [
|
| 80 |
+
"fsdp"
|
| 81 |
+
],
|
| 82 |
+
"model.norm.weight": [
|
| 83 |
+
"fsdp"
|
| 84 |
+
],
|
| 85 |
+
"model.layers.*": [
|
| 86 |
+
[
|
| 87 |
+
"data",
|
| 88 |
+
"fsdp"
|
| 89 |
+
],
|
| 90 |
+
null,
|
| 91 |
+
null
|
| 92 |
+
],
|
| 93 |
+
"lm_head": [
|
| 94 |
+
[
|
| 95 |
+
"data",
|
| 96 |
+
"fsdp"
|
| 97 |
+
],
|
| 98 |
+
null,
|
| 99 |
+
null
|
| 100 |
+
]
|
| 101 |
+
}
|
| 102 |
+
}
|
000000020000/model.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:369764e76e0440c8b6915363e5c05c56c273a7e5012b6722f46f5083e2cb0d18
|
| 3 |
+
size 7248226091
|