Nov 2 2 days ago 01:19 Add native int8 HQ quantization of Beta7 Fix1(INT8 ConvRot + protected edges) This PR adds a native quantized HQ checkpoint of 10Eros_Max_h3_8stepTURBO_beta7_fix1.
It uses native INT8 ConvRot for the inner 180 main transformer matrices and native precision on protected edge blocks (0,1,47-49), preserving all other auxiliary tensors in their original source precision.
This is the fixed native int8 recipe, a quality-focused HQ activation learned w6a8+int8 version will follow for lower memory requirements and lower-end GPUs in my own repo.
Actual VRAM usage depends on the workflow and other loaded models.
A full model card and standalone release will follow. Nov 1 3 days ago 22:47 Upload 10Eros_Max_h3_8stepTURBO_beta7_fix1.safetensors Oct 30 5 days ago 03:29 Update README.md 03:29 Update README.md 01:53 Upload 10Eros_Max_h3_raw_beta7_int8.safetensors Oct 28 7 days ago 07:16 Update README.md Oct 27 8 days ago 23:40 Upload 10Eros_Max_h3_8stepTURBO_beta7.safetensors 23:22 Update README.md 23:18 Update README.md 23:17 Update README.md 22:52 Upload 10Eros_Max_h3_8stepTURBO_beta7_int8HQ.safetensors Oct 13 22 days ago 09:03 Update README.md 09:03 Update README.md Oct 11 24 days ago 07:45 Upload Loras/Eros_beta6mix_Lora.safetensors Oct 9 26 days ago 21:52 Upload 10Eros_Max_FastH3_VSAgates_UNSTABLE_beta6-int8_HQ.safetensors 07:13 Update README.md 07:13 Update README.md 07:10 Rename beta6_sampling_example to beta6_sampling_example.png 07:10 Upload beta6_sampling_example 06:45 Update README.md 06:23 Update README.md 05:55 Upload 3 files Sep 8 21:42 change typo (multires) in README.md (#42 ) TenStrip +1: wittywolk - Update README.md (95139d379378564ccbf2ac90ada4d72ab65ff6f5)
Co-authored-by: Robert <wittywolk@users.noreply.huggingface.co> 21:42 Add Non-Turbo optimized w4a8 version (#41 ) TenStrip +1: LokkenJP - Add Non-Turbo optimized w4a8 version (9251704c14150f5d6d60fbdd4d8b7ad93e0ec465)
Co-authored-by: George Well <LokkenJP@users.noreply.huggingface.co> Sep 7 20:18 Update README.md 20:17 Update README.md 20:16 Update README.md 20:16 Update README.md 20:15 Update README.md 20:02 Upload 5 files Aug 30 01:20 Upload 2 files Aug 28 02:50 Update README.md 02:49 Update README.md 02:48 Update README.md 02:47 Update README.md Aug 25 19:08 Update README.md 06:06 Delete 10Eros_Max_h3_fl2va_beta1_pruned.safetensors 06:06 Delete 10Eros_Max_h3_fl2va_bf16_test3_pruned.safetensors 06:06 Delete 10Eros_Max_h3_fl2va_bf16_test4_pruned.safetensors 06:02 Upload 2 files Aug 19 05:11 Update README.md 04:36 Update README.md 04:33 Upload 10Eros_Max_h3_TURBO_ref2va_beta2.safetensors Aug 18 21:30 Update README.md 18:35 Upload 10Eros_Max_h3_ref2va_beta2_pruned.safetensors 17:55 Delete 10Eros_Max_h3_fl2va_bf16_test2_pruned.safetensors 17:24 Upload h3_graft_methodology.md 17:23 Update README.md Aug 14 19:25 Upload 10Eros_Max_h3_fl2va_beta2_pruned.safetensors Aug 12 04:29 Update README.md 
