SearchingMan commited on
Commit
d1c673a
·
verified ·
1 Parent(s): f8f6eec

Document BF16 INT8 conditioning cosine parity

Browse files
Files changed (1) hide show
  1. README.md +12 -0
README.md CHANGED
@@ -69,6 +69,18 @@ Download these from [Comfy-Org/MiniMax-H3](https://huggingface.co/Comfy-Org/Mini
69
 
70
  The showcase used one matched 1344x768, 243-frame, 24 FPS, 25-step T2V contract. The three source outputs and merged comparison are technical assets. They do not establish a visual-quality winner. The recovered 8B showcase uses the BF16 base; the INT8 ConvRot variant has separate numerical conditioning-parity evidence.
71
 
 
 
 
 
 
 
 
 
 
 
 
 
72
  The requested hard cuts at 3 and 7 seconds did not pass for any encoder. Read `evidence/cut_diagnostics.json` before using the showcase as prompt-adherence evidence.
73
 
74
  The recovered 8B and pruned-24 variants were validated for text-only T2V. Do not use them for image, first-frame, last-frame, or reference inputs.
 
69
 
70
  The showcase used one matched 1344x768, 243-frame, 24 FPS, 25-step T2V contract. The three source outputs and merged comparison are technical assets. They do not establish a visual-quality winner. The recovered 8B showcase uses the BF16 base; the INT8 ConvRot variant has separate numerical conditioning-parity evidence.
71
 
72
+ ### Recovered 8B BF16 versus INT8 ConvRot conditioning parity
73
+
74
+ For one fixed seven-token prompt, the recovered 8B BF16 and INT8 ConvRot variants produced conditioning tensors with shape `[1, 7, 5120]`. The output tensors were flattened and compared with cosine similarity:
75
+
76
+ | Metric | Result |
77
+ |---|---:|
78
+ | Cosine similarity | `0.9999650716781616` |
79
+ | Mean absolute difference | `0.015660330653190613` |
80
+ | Maximum absolute difference | `0.6452102661132812` |
81
+
82
+ This is a numerical conditioning-parity diagnostic for one prompt. It is not a visual-quality score, a prompt-adherence score, or proof that the INT8 variant produces identical video results. Full details are in `evidence/release_loader_smoke.json`.
83
+
84
  The requested hard cuts at 3 and 7 seconds did not pass for any encoder. Read `evidence/cut_diagnostics.json` before using the showcase as prompt-adherence evidence.
85
 
86
  The recovered 8B and pruned-24 variants were validated for text-only T2V. Do not use them for image, first-frame, last-frame, or reference inputs.