mirror of
https://github.com/deepseek-ai/DeepInfer.git
synced 2026-10-28 23:07:09 +08:00
8a5209ee65
- examples/neuron: tiny random V4.1 checkpoint in the real HF layout, pure-torch replacements for the official reference kernels, golden generation, a teacher-forced DeepInfer runner (LLM offline API) and a comparison tool. CPU mode: 25/27 greedy tokens match, worst logits cos 0.9985 / rel 0.056 against a reference self-noise floor of 0.032-0.060. - model_loader: check parameters against the platform device, not is_cuda. - deep_gemm paged MQA logits: do not dereference block-table slots past the context. - bind_cpu: skip in Neuron CPU mode (1 core per NeuronCore share would serialize host compute). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
196 lines
2.1 KiB
JSON
196 lines
2.1 KiB
JSON
{
|
|
"name": "short",
|
|
"prompt_text": "The quick brown fox jumps over the lazy dog.",
|
|
"prompt_ids": [
|
|
0,
|
|
671,
|
|
4787,
|
|
13769,
|
|
46012,
|
|
54994,
|
|
1060,
|
|
270,
|
|
41638,
|
|
6397,
|
|
16
|
|
],
|
|
"n_prompt_tokens": 11,
|
|
"tokens": [
|
|
85792,
|
|
78870,
|
|
110111,
|
|
9689,
|
|
101353,
|
|
107157,
|
|
84220,
|
|
19174,
|
|
82767
|
|
],
|
|
"decoded": "\u6301\u3061))\n\n\ncombined Pract@media acuity\u7ea4\u7ef4\u7d20 Sin\u89c4\u8303\u7684",
|
|
"top1_top2_margin": [
|
|
0.10489,
|
|
0.18469,
|
|
0.09769,
|
|
0.10715,
|
|
0.06664,
|
|
0.23189,
|
|
0.33355,
|
|
0.29129,
|
|
0.39137
|
|
],
|
|
"top5_ids": [
|
|
[
|
|
85792,
|
|
109217,
|
|
110226,
|
|
91841,
|
|
87149
|
|
],
|
|
[
|
|
78870,
|
|
62985,
|
|
4375,
|
|
37823,
|
|
71835
|
|
],
|
|
[
|
|
110111,
|
|
62076,
|
|
14882,
|
|
127459,
|
|
10901
|
|
],
|
|
[
|
|
9689,
|
|
75659,
|
|
30795,
|
|
6070,
|
|
49485
|
|
],
|
|
[
|
|
101353,
|
|
46574,
|
|
36355,
|
|
20609,
|
|
21109
|
|
],
|
|
[
|
|
107157,
|
|
5900,
|
|
31951,
|
|
75048,
|
|
62107
|
|
],
|
|
[
|
|
84220,
|
|
16903,
|
|
26762,
|
|
104561,
|
|
106094
|
|
],
|
|
[
|
|
19174,
|
|
94517,
|
|
53537,
|
|
46354,
|
|
43228
|
|
],
|
|
[
|
|
82767,
|
|
70793,
|
|
124801,
|
|
69604,
|
|
40493
|
|
]
|
|
],
|
|
"top5_logits": [
|
|
[
|
|
4.30327,
|
|
4.19838,
|
|
4.06756,
|
|
4.05694,
|
|
4.0096
|
|
],
|
|
[
|
|
4.72153,
|
|
4.53684,
|
|
4.06865,
|
|
4.00015,
|
|
3.96316
|
|
],
|
|
[
|
|
4.20809,
|
|
4.1104,
|
|
4.08111,
|
|
4.05279,
|
|
4.05153
|
|
],
|
|
[
|
|
4.17989,
|
|
4.07274,
|
|
3.99208,
|
|
3.97799,
|
|
3.95346
|
|
],
|
|
[
|
|
4.39166,
|
|
4.32503,
|
|
3.96568,
|
|
3.95226,
|
|
3.93489
|
|
],
|
|
[
|
|
4.5328,
|
|
4.30091,
|
|
4.14662,
|
|
4.01601,
|
|
3.98233
|
|
],
|
|
[
|
|
4.48246,
|
|
4.14891,
|
|
4.05826,
|
|
4.0559,
|
|
4.02218
|
|
],
|
|
[
|
|
4.82395,
|
|
4.53265,
|
|
4.26447,
|
|
4.00352,
|
|
3.94851
|
|
],
|
|
[
|
|
4.66934,
|
|
4.27798,
|
|
3.91097,
|
|
3.8789,
|
|
3.85056
|
|
]
|
|
],
|
|
"logits_l2": [
|
|
359.5489,
|
|
359.0275,
|
|
360.0101,
|
|
359.5485,
|
|
358.1219,
|
|
360.3374,
|
|
359.3568,
|
|
358.4936,
|
|
358.7865
|
|
],
|
|
"logits_mean": [
|
|
0.002626,
|
|
0.004761,
|
|
-0.003585,
|
|
0.002637,
|
|
-0.00203,
|
|
-0.000569,
|
|
0.000388,
|
|
-0.001356,
|
|
0.003071
|
|
],
|
|
"full_golden": "/home/ubuntu/di/golden/short.pt",
|
|
"full_golden_sha256": "59f13484afa07472500d07958212c54203e7bd99b89de3fb936e10af93deb9a6",
|
|
"seconds": 3.6
|
|
} |