remove v75 bundle: not Plane-B-validated (Plane B load-fail (6GB > ~4.7GB DSP ceiling; W4 HTP-blocked))
Browse files- v75/qwen3.5-4b-1024.json +0 -69
- v75/qwen354b_decode_s0_w8.bin +0 -3
- v75/qwen354b_decode_s1_w8.bin +0 -3
- v75/qwen354b_embed_f16.bin +0 -3
- v75/qwen354b_lmhead_f16.bin +0 -3
- v75/tokenizer.json +0 -3
v75/qwen3.5-4b-1024.json
DELETED
|
@@ -1,69 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"schema_version": 1,
|
| 3 |
-
"model": {
|
| 4 |
-
"name": "qwen3.5-4b-1024",
|
| 5 |
-
"family": "llm",
|
| 6 |
-
"dsp_arch": "v75",
|
| 7 |
-
"tokenizer_pre": "qwen2"
|
| 8 |
-
},
|
| 9 |
-
"params": {
|
| 10 |
-
"hidden": 2560,
|
| 11 |
-
"vocab": 248320,
|
| 12 |
-
"n_layers": 32,
|
| 13 |
-
"max_ctx": 1024,
|
| 14 |
-
"kv_dim": 1024,
|
| 15 |
-
"head_dim": 256,
|
| 16 |
-
"rope_theta": 10000000.0,
|
| 17 |
-
"eos_token_id": 248044,
|
| 18 |
-
"gated_delta": {
|
| 19 |
-
"q_heads": 16,
|
| 20 |
-
"kv_heads": 4,
|
| 21 |
-
"lin_k_heads": 16,
|
| 22 |
-
"lin_v_heads": 32,
|
| 23 |
-
"lin_k_dim": 128,
|
| 24 |
-
"lin_v_dim": 128,
|
| 25 |
-
"conv_kernel": 4,
|
| 26 |
-
"full_every": 4,
|
| 27 |
-
"rot_dim": 64,
|
| 28 |
-
"attn_gate": true,
|
| 29 |
-
"full_attn_idx": [
|
| 30 |
-
3,
|
| 31 |
-
7,
|
| 32 |
-
11,
|
| 33 |
-
15,
|
| 34 |
-
19,
|
| 35 |
-
23,
|
| 36 |
-
27,
|
| 37 |
-
31
|
| 38 |
-
]
|
| 39 |
-
}
|
| 40 |
-
},
|
| 41 |
-
"artifacts": {
|
| 42 |
-
"contexts": {
|
| 43 |
-
"decode_s0": {
|
| 44 |
-
"bin": "qwen354b_decode_s0_w8.bin"
|
| 45 |
-
},
|
| 46 |
-
"decode_s1": {
|
| 47 |
-
"bin": "qwen354b_decode_s1_w8.bin"
|
| 48 |
-
},
|
| 49 |
-
"lmhead": {
|
| 50 |
-
"bin": "qwen354b_lmhead_f16.bin"
|
| 51 |
-
}
|
| 52 |
-
},
|
| 53 |
-
"embed": "qwen354b_embed_f16.bin",
|
| 54 |
-
"tokenizer": "tokenizer.json"
|
| 55 |
-
},
|
| 56 |
-
"plan": {
|
| 57 |
-
"steps": [
|
| 58 |
-
{
|
| 59 |
-
"host": "qwen3_5_generate",
|
| 60 |
-
"params": {
|
| 61 |
-
"decode": "qwen354b_decode_s0_w8,qwen354b_decode_s1_w8",
|
| 62 |
-
"lmhead": "qwen354b_lmhead_f16",
|
| 63 |
-
"max_new": 64,
|
| 64 |
-
"neg": -50000
|
| 65 |
-
}
|
| 66 |
-
}
|
| 67 |
-
]
|
| 68 |
-
}
|
| 69 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
v75/qwen354b_decode_s0_w8.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:d0866ab249efd39cc44a05bffeb4e46eb7b814403f6d6fd2976eef7c27a78db0
|
| 3 |
-
size 1804955648
|
|
|
|
|
|
|
|
|
|
|
|
v75/qwen354b_decode_s1_w8.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:f57603c60fdb78e0ea341179e01bab89e5676d69f01847d7b90323c6cc7d02eb
|
| 3 |
-
size 1804959744
|
|
|
|
|
|
|
|
|
|
|
|
v75/qwen354b_embed_f16.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:c78988d979f52a340f39cd69edd8e4b0a80218a5fa8d95b6158bfce4a3de2a34
|
| 3 |
-
size 1271398400
|
|
|
|
|
|
|
|
|
|
|
|
v75/qwen354b_lmhead_f16.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:4b82afa1d43bcbaa99d429807e9f05c2d97e26baa8a4f41d878276f9ab44fb43
|
| 3 |
-
size 1272709120
|
|
|
|
|
|
|
|
|
|
|
|
v75/tokenizer.json
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:06b9509352d2af50381ab2247e083b80d32d5c0aba91c272ca9ff729b6a0e523
|
| 3 |
-
size 19989325
|
|
|
|
|
|
|
|
|
|
|
|