Upload folder using huggingface_hub
Browse files- 000000000827/config.json +66 -0
- 000000000827/model.pt +3 -0
000000000827/config.json
ADDED
|
@@ -0,0 +1,66 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"type": "oloop.OLoopModel",
|
| 3 |
+
"pretrained_url": "aklein4/Llama-3.2-1B-TPU",
|
| 4 |
+
"pretrained_step": 0,
|
| 5 |
+
"pretrained_strict": false,
|
| 6 |
+
"torch_dtype": "float32",
|
| 7 |
+
"vocab_size": 128256,
|
| 8 |
+
"bos_token_id": 128000,
|
| 9 |
+
"eos_token_id": 128001,
|
| 10 |
+
"pad_token_id": -100,
|
| 11 |
+
"hidden_size": 2048,
|
| 12 |
+
"num_hidden_layers": 16,
|
| 13 |
+
"num_attention_heads": 32,
|
| 14 |
+
"num_key_value_heads": 8,
|
| 15 |
+
"intermediate_size": 8192,
|
| 16 |
+
"hidden_act": "silu",
|
| 17 |
+
"max_position_embeddings": 131072,
|
| 18 |
+
"rope_theta": 500000.0,
|
| 19 |
+
"rope_scaling": {
|
| 20 |
+
"factor": 32.0,
|
| 21 |
+
"high_freq_factor": 4.0,
|
| 22 |
+
"low_freq_factor": 1.0,
|
| 23 |
+
"original_context_len": 8192
|
| 24 |
+
},
|
| 25 |
+
"initializer_range": 0.02,
|
| 26 |
+
"attention_dropout": false,
|
| 27 |
+
"attention_bias": false,
|
| 28 |
+
"rms_norm_eps": 1e-05,
|
| 29 |
+
"attention_kernel": "flash_attention",
|
| 30 |
+
"pure_modules": [],
|
| 31 |
+
"fast_weight_size": 2048,
|
| 32 |
+
"base_lr": 0.0,
|
| 33 |
+
"momentum_beta": 0.9,
|
| 34 |
+
"momentum_dtype": "bfloat16",
|
| 35 |
+
"state_dtype": "float32",
|
| 36 |
+
"sharding": {
|
| 37 |
+
"model.layers.*": [
|
| 38 |
+
[
|
| 39 |
+
"data",
|
| 40 |
+
"fsdp"
|
| 41 |
+
],
|
| 42 |
+
null,
|
| 43 |
+
null
|
| 44 |
+
],
|
| 45 |
+
"lm_head": [
|
| 46 |
+
[
|
| 47 |
+
"data",
|
| 48 |
+
"fsdp"
|
| 49 |
+
],
|
| 50 |
+
null,
|
| 51 |
+
null
|
| 52 |
+
]
|
| 53 |
+
},
|
| 54 |
+
"remat": {
|
| 55 |
+
"activation_checkpoint_layers": [
|
| 56 |
+
"LlamaDecoderLayer"
|
| 57 |
+
],
|
| 58 |
+
"optimization_barrier_layers": [
|
| 59 |
+
"LlamaDecoderLayer"
|
| 60 |
+
],
|
| 61 |
+
"scan_layers": "model.layers",
|
| 62 |
+
"offload_tensors": [
|
| 63 |
+
"decoder_input"
|
| 64 |
+
]
|
| 65 |
+
}
|
| 66 |
+
}
|
000000000827/model.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:65f82194570066f1c9d2844ae56d2fbb9b35b3f1566bb6991dda03aff5e1a106
|
| 3 |
+
size 7067733759
|