Upload folder using huggingface_hub
Browse files
snake-45653760-a024c281-checkpoint-0/config.h
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
#pragma once
|
| 2 |
+
|
| 3 |
+
#define N_LAYERS 12
|
| 4 |
+
#define VOCAB_SIZE 256
|
| 5 |
+
#define D_MODEL 768
|
| 6 |
+
#define D_INNER 1536
|
| 7 |
+
#define DT_RANK 48
|
| 8 |
+
#define D_STATE 16
|
| 9 |
+
#define D_CONV 4
|
snake-45653760-a024c281-checkpoint-0/config.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"d_model": 768,
|
| 3 |
+
"d_intermediate": 0,
|
| 4 |
+
"n_layer": 12,
|
| 5 |
+
"vocab_size": 256,
|
| 6 |
+
"ssm_cfg": {},
|
| 7 |
+
"attn_layer_idx": [],
|
| 8 |
+
"attn_cfg": {},
|
| 9 |
+
"rms_norm": true,
|
| 10 |
+
"residual_in_fp32": true,
|
| 11 |
+
"fused_add_norm": true,
|
| 12 |
+
"pad_vocab_size_multiple": 8,
|
| 13 |
+
"tie_embeddings": false
|
| 14 |
+
}
|
snake-45653760-a024c281-checkpoint-0/model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dce42b36b291dfdb11ab0f87329dd1ad11d9a87386c3160b02088c1bec924e77
|
| 3 |
+
size 182615040
|
snake-45653760-a024c281-checkpoint-0/pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:94d7d355f2ba83d8d6e50ff90e37de68a500cd3f55623cdaf2869e9f6df7f325
|
| 3 |
+
size 182661686
|
snake-45653760-a024c281-checkpoint-0/training_params.json
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"d_model": 768,
|
| 3 |
+
"n_layers": 12,
|
| 4 |
+
"vocab_size": 256,
|
| 5 |
+
"model_size": 45653760,
|
| 6 |
+
"learning_rate": 0.001,
|
| 7 |
+
"seq_length": 1024,
|
| 8 |
+
"batch_size": 4,
|
| 9 |
+
"num_batches": 229,
|
| 10 |
+
"tokenizer_path": "",
|
| 11 |
+
"dataset_path": "flpelerin/tinystories-1k",
|
| 12 |
+
"model_name": "snake-45653760-a024c281",
|
| 13 |
+
"wandb_project": "snake-training",
|
| 14 |
+
"wandb_user": "florianpelerin110304",
|
| 15 |
+
"hf_repo": "flpelerin/snakes",
|
| 16 |
+
"target_epoch": 10,
|
| 17 |
+
"current_epoch": 0
|
| 18 |
+
}
|