Add whisper-base v75 bundle (Galaxy S24 / Hexagon v75)
#3
by flying101 - opened
- v75/tokenizer.json +0 -0
- v75/whisper-base.json +46 -0
- v75/whisper_base_mel_filters.bin +3 -0
- v75/whisper_conv_stem.bin +3 -0
- v75/whisperbase_dec_f16.bin +3 -0
- v75/whisperbase_enc_f16.bin +3 -0
v75/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
v75/whisper-base.json
ADDED
|
@@ -0,0 +1,46 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"schema_version": 1,
|
| 3 |
+
"_comment": "Whisper-base ASR (encoder-decoder). Host log-mel + host conv stem (the conv is HTP-hostile in-graph) feed the encoder graph (input encoder_features); the autoregressive decoder (causal self-attention + cross-attention to the cached encoder states, learned positional embeddings, tied lm-head) emits tokens, detok'd on the host. Gated on WER vs HF.",
|
| 4 |
+
"model": {
|
| 5 |
+
"name": "whisper-base",
|
| 6 |
+
"family": "asr",
|
| 7 |
+
"dsp_arch": "v75",
|
| 8 |
+
"tokenizer_pre": "whisper"
|
| 9 |
+
},
|
| 10 |
+
"params": {
|
| 11 |
+
"hidden": 512,
|
| 12 |
+
"vocab": 51865,
|
| 13 |
+
"n_layers": 6,
|
| 14 |
+
"max_ctx": 200,
|
| 15 |
+
"kv_dim": 512,
|
| 16 |
+
"head_dim": 64,
|
| 17 |
+
"rope_theta": 10000.0,
|
| 18 |
+
"eos_token_id": 50257
|
| 19 |
+
},
|
| 20 |
+
"artifacts": {
|
| 21 |
+
"contexts": {
|
| 22 |
+
"encoder": {
|
| 23 |
+
"bin": "whisperbase_enc_f16.bin"
|
| 24 |
+
},
|
| 25 |
+
"decoder": {
|
| 26 |
+
"bin": "whisperbase_dec_f16.bin"
|
| 27 |
+
}
|
| 28 |
+
},
|
| 29 |
+
"tokenizer": "tokenizer.json"
|
| 30 |
+
},
|
| 31 |
+
"plan": {
|
| 32 |
+
"steps": [
|
| 33 |
+
{
|
| 34 |
+
"host": "asr_transcribe",
|
| 35 |
+
"params": {
|
| 36 |
+
"encoder": "whisperbase_enc_f16",
|
| 37 |
+
"decoder": "whisperbase_dec_f16",
|
| 38 |
+
"sot": 50258,
|
| 39 |
+
"mel_filters": "whisper_base_mel_filters.bin",
|
| 40 |
+
"conv_weights": "whisper_conv_stem.bin",
|
| 41 |
+
"n_mels": 80
|
| 42 |
+
}
|
| 43 |
+
}
|
| 44 |
+
]
|
| 45 |
+
}
|
| 46 |
+
}
|
v75/whisper_base_mel_filters.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d18b48f33345d958ea71098bb990b5e32cebc78d9a2cf65ca0d7302f34ebe8a9
|
| 3 |
+
size 64320
|
v75/whisper_conv_stem.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ab32f7839de3b8db4e4eed32b743f774fce82b00b0a2a414708272a0971a2f26
|
| 3 |
+
size 6713344
|
v75/whisperbase_dec_f16.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:55fba1c66983b9b4e081cb4a978811c44c65bcb6830efc40bf1195dfa292ce7b
|
| 3 |
+
size 152276992
|
v75/whisperbase_enc_f16.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5a82c953b6f2c26649985a8aa29a557de2b9b3c2b0eb6f7a43e21cca3c2ff146
|
| 3 |
+
size 56061952
|