Correct the licence documentation

NOTICE

@@ -6,10 +6,17 @@ Unsloth has not modified the weights: every file here is byte for byte the file
 
 The weights derive from Wan2.2-TI2V-5B (https://huggingface.co/Wan-AI/Wan2.2-TI2V-5B),
 Copyright 2024-2025 The Alibaba Wan Team Authors, licensed under the Apache License,
-Version 2.0. The full licence text is in LICENSE.
+Version 2.0. LICENSE is the text the Wan team ships at
+https://github.com/Wan-Video/Wan2.2/blob/main/LICENSE.txt, verbatim, appendix placeholder
+included. Upstream ships no NOTICE file, so there is nothing further to propagate under
+Apache-2.0 section 4(d).
 
-Changes relative to Wan-AI/Wan2.2-TI2V-5B (Apache-2.0 section 4(b)): the transformer has
-been quantised to GGUF at 14 precisions (Q2_K through Q8_0), which changes the numerics,
-and the VAE is shipped as one standalone safetensors file. Those changes are QuantStack's
-work, not Unsloth's. Not an official Alibaba Wan Team or QuantStack product, and not
-endorsed by either.
+Changes relative to Wan-AI/Wan2.2-TI2V-5B (Apache-2.0 section 4(b)), all of them
+QuantStack's work and none of them Unsloth's:
+
+  * the diffusion transformer was quantised to GGUF at 13 precisions (Q2_K, Q3_K_S, Q3_K_M,
+    Q4_0, Q4_1, Q4_K_S, Q4_K_M, Q5_0, Q5_1, Q5_K_S, Q5_K_M, Q6_K, Q8_0), which changes the
+    numerics;
+  * the VAE was converted from the original Wan2.2_VAE.pth to Wan2.2_VAE.safetensors.
+
+Not an official Alibaba Wan Team or QuantStack product, and not endorsed by either.