Cherry pick v1.24.0#2439

Open

xin3he wants to merge 31 commits intomasterfrom

cherry_pick_v1.24.0

Commits on Mar 31, 2026

[SW-240730] Support Compressed Tensors quantization method with fp8 w… (#303)

Yantom1
authored and
xinhe3
committed
[SW-240400] Fix MoE weights handling in measure (#315)

ulivne
authored and
xinhe3
committed
[SW-240869] add support for str padding - same and valid (#311)

linoybu
authored and
xinhe3
committed
[PERFC-756] skip xpu qunaitzed func wrapper test (#316)

ulivne
authored and
xinhe3
committed
corret dequant func check (#318)

authored and
xinhe3
committed
[PERFC-756] fix skipping test (#320)

ulivne
authored and
xinhe3
committed
[SW-240800] Add option to specify output tensor in torch.matmul (#306)

authored and
xinhe3
committed
[SW-233758] Support dynamic quantization for Matmul (#317)

nirda7
authored and
xinhe3
committed
[SW-233758] Adjust Matmul axis for scale calculation according to input (#323)

nirda7
authored and
xinhe3
committed
[GAUDISW-5809] - Distinguish runtime scale patching from dynamic quan… (#327)

nirda7
authored and
xinhe3
committed
[GAUDISW-244631] dispatch quantized hidden_states (#337)

xinyu-intel
authored and
xinhe3
committed
[GAUDISW-228042] Add support for dynamic vLLM kv-cache quantization (#330)

dudilester
authored and
xinhe3
committed
[GAUDISW-244192] - Set whether using dynamic quantization from Inc (#326)

nirda7
authored and
xinhe3
committed
disable autoround tests [GAUDISW-245272] (#342)

linoybu
authored and
xinhe3
committed
[GAUDISW-245117] add b2b op (#341)

linoybu
authored and
xinhe3
committed
[GAUDISW-244752] add dynamic scale for V-Cache on Hiddden dim (#339)

dudilester
authored and
xinhe3
committed
[GAUDISW-245131] Skip test for load/save model checkpoint (#345)

authored and
xinhe3
committed
[GAUDISW-244752] Fix torch.compile graph break in v-cache hidden scale update (#350)

dudilester
authored and
xinhe3
committed
[GAUDISW-245917] Added WA for weights in 3d - Granite 4.0 (#353)

HolyFalafel
authored and
xinhe3
committed
[GAUDISW-245950] disable test fp8_aware_gptq (#352)

linoybu
authored and
xinhe3
committed
[GAUDISW-245131] Skip test for load/save model checkpoint part 2 (#351)

Silv3S
authored and
xinhe3
committed
[GAUDISW-224538] Calling init_linear in __init__ (#357)

HolyFalafel
authored and
xinhe3
committed
[GAUDISW-246337] Added dynamic quant with weight PCS POW2 (#354)

HolyFalafel
authored and
xinhe3
committed
[GAUDISW-246352] assign value of 1 to scales of non-active blocks (#358)

dudilester
authored and
xinhe3
committed
[GAUDISW-246550] Remove spaces before equals in scale_method_config parameter assignments (#362)

authored and
xinhe3
committed
[GAUDISW-246083] - adjust load mode to pcs for new pytorch version (#363)

nirda7
authored and
xinhe3
committed
[GAUDISW-247053] Fix coverity issues take2 (#369)

authored and
xinhe3
committed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Cherry pick v1.24.0#2439

Cherry pick v1.24.0#2439
xin3he wants to merge 31 commits intomasterfrom
cherry_pick_v1.24.0

Commits on Mar 31, 2026

[SW-240730] Support Compressed Tensors quantization method with fp8 w… (#303)

[SW-240400] Fix MoE weights handling in measure (#315)

[SW-240869] add support for str padding - same and valid (#311)

[PERFC-756] skip xpu qunaitzed func wrapper test (#316)

corret dequant func check (#318)

[PERFC-756] fix skipping test (#320)

[SW-240800] Add option to specify output tensor in torch.matmul (#306)

[SW-233758] Support dynamic quantization for Matmul (#317)

[SW-233758] Adjust Matmul axis for scale calculation according to input (#323)

[GAUDISW-5809] - Distinguish runtime scale patching from dynamic quan… (#327)

[GAUDISW-244631] dispatch quantized hidden_states (#337)

[GAUDISW-228042] Add support for dynamic vLLM kv-cache quantization (#330)

[GAUDISW-244192] - Set whether using dynamic quantization from Inc (#326)

disable autoround tests [GAUDISW-245272] (#342)

[GAUDISW-245117] add b2b op (#341)

[GAUDISW-244752] add dynamic scale for V-Cache on Hiddden dim (#339)

[GAUDISW-245131] Skip test for load/save model checkpoint (#345)

[GAUDISW-244752] Fix torch.compile graph break in v-cache hidden scale update (#350)

[GAUDISW-245917] Added WA for weights in 3d - Granite 4.0 (#353)

[GAUDISW-245950] disable test fp8_aware_gptq (#352)

[GAUDISW-245131] Skip test for load/save model checkpoint part 2 (#351)

[GAUDISW-224538] Calling init_linear in init (#357)

[GAUDISW-246337] Added dynamic quant with weight PCS POW2 (#354)

[GAUDISW-246352] assign value of 1 to scales of non-active blocks (#358)

[GAUDISW-246550] Remove spaces before equals in scale_method_config parameter assignments (#362)

[GAUDISW-246083] - adjust load mode to pcs for new pytorch version (#363)

[GAUDISW-247053] Fix coverity issues take2 (#369)

Commits on Apr 20, 2026

add missing changes

Commits on Apr 29, 2026

merge branch master and fix bug

[pre-commit.ci] auto fixes from pre-commit.com hooks

Commits on May 8, 2026

Merge branch 'master' into cherry_pick_v1.24.0