topabaem commited on
Commit
3506ed7
·
verified ·
1 Parent(s): fa92dce

Play the samples in the card instead of linking them

Browse files
Files changed (1) hide show
  1. README.md +20 -1
README.md CHANGED
@@ -96,8 +96,27 @@ deserves is that neither build followed that prompt — it asked for an astronau
96
  and got neither — so the model had no confident answer there for a small
97
  conditioning change to disturb.
98
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
99
  `samples/README.md` carries the prompts, the per-clip conditioning drift, and
100
- what these clips do and do not establish.
 
101
 
102
  ## Memory
103
 
 
96
  and got neither — so the model had no confident answer there for a small
97
  conditioning change to disturb.
98
 
99
+ ### Watch the two that matter
100
+
101
+ **Forge — the pair holding together.** Deterministic sampler, same seed. Left is
102
+ BF16, right is 4-bit.
103
+
104
+ <video controls width="100%" src="https://huggingface.co/topabaem/LTX-2.5-Text-Encoder-4bit/resolve/main/samples/compare-det-02.mp4"></video>
105
+
106
+ **Dune — the pair that does not.** Same terrain, same sun, same shadow, same
107
+ walk, and a different person.
108
+
109
+ <video controls width="100%" src="https://huggingface.co/topabaem/LTX-2.5-Text-Encoder-4bit/resolve/main/samples/compare-det-01.mp4"></video>
110
+
111
+ Individually, the forge under the vendor's own sampler:
112
+
113
+ | BF16 original | 4-bit |
114
+ |---|---|
115
+ | <video controls width="100%" src="https://huggingface.co/topabaem/LTX-2.5-Text-Encoder-4bit/resolve/main/samples/bf16-02.mp4"></video> | <video controls width="100%" src="https://huggingface.co/topabaem/LTX-2.5-Text-Encoder-4bit/resolve/main/samples/4bit-02.mp4"></video> |
116
+
117
  `samples/README.md` carries the prompts, the per-clip conditioning drift, and
118
+ what these clips do and do not establish. All thirty clips are in `samples/`,
119
+ and the Space plays them side by side under its **BF16 vs 4-bit** tab.
120
 
121
  ## Memory
122