Hugging Face's logo

slvDev
/
esp32-ai-tinystories

Text Generation
English
esp32
esp32-s3
microcontroller
embedded
edge
per-layer-embeddings
tinystories
int4
Model card Files
xet
 Community
esp32-ai-tinystories
File size: 2,388 Bytes
6974e24
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
{
  "architecture": "PLE",
  "format": {
    "magic_ascii": "PLE\\0",
    "magic_le_hex": "0x00454c50",
    "version": 1,
    "header_bytes": 56,
    "flags": {
      "TIED_HEAD": true
    }
  },
  "config": {
    "input_vocab": 32768,
    "output_vocab": 25353,
    "active_tokenizer_entries": 25353,
    "d_model": 96,
    "n_layers": 6,
    "n_heads": 4,
    "ffn_hidden": 66,
    "ple_dim": 128,
    "seq_len": 256,
    "rope_theta": 10000.0
  },
  "parameters": {
    "ple_table": 25165824,
    "tok_emb": 3145728,
    "core": 556416,
    "total": 28867968
  },
  "quantization": {
    "weights": "int4",
    "group_size": 128
  },
  "runtime": {
    "weights_staged_int8": "per-position core and tied head, in PSRAM",
    "activations": "int8, quantized per staged matvec",
    "sram": "float scratch buffers and RMSNorm vectors",
    "psram": "staged core, staged head, KV cache, logits",
    "flash": "PLE table and token embedding, memory-mapped"
  },
  "measured": {
    "as_published": {
      "ms_per_token": 102.9,
      "tok_per_second": 9.5,
      "date": "2026-07-21"
    },
    "current": {
      "ms_per_token": 94.9,
      "tok_per_second": 9.88,
      "staged_tensors": 44,
      "managed_sram_bytes": 29320,
      "psram_mb": 4.19,
      "build": "-O3"
    }
  },
  "device_fingerprint_fnv1a": "a9bdd778",
  "files": {
    "model.bin": {
      "sha256": "1d8326c05c383ccfa615f5455575802817cb453dbc7ab28875d41a9dbb45477e",
      "bytes": 14912348,
      "license": "MIT"
    },
    "tokenizer.json": {
      "sha256": "4e28163669f2249af31a528a54fc25064dcbd0a34edbfa7bedb16d2d600ec7ae",
      "bytes": 1788896,
      "license": "MIT"
    }
  },
  "weights_license": "MIT",
  "tokenizer_license": "MIT",
  "training_data_license": "CDLA-Sharing-1.0",
  "source_checkpoint": {
    "name": "ple-cleandeploy-s0",
    "distributed": false,
    "sha256": "746950c43e85c6e42e0cdaaf7bee72100e70be40a0bbaa473d57fbf85e17600a"
  },
  "training_data": {
    "dataset": "roneneldan/TinyStories",
    "slice_bytes": 314572800,
    "revision_pinned": false,
    "redistributed": false,
    "note": "prepare.py fetches the dataset's main branch and records no hash of the raw slice, so a re-run reproduces the method, not the bytes"
  },
  "hardware": {
    "board": "ESP32-S3",
    "flash_mb": 16,
    "psram_mb": 8,
    "sram_kb": 512,
    "model_partition_offset": "0x110000"
  }
}