[staging CI] unslothai/unsloth#6935 - #281

Closed

danielhanchen wants to merge 40 commits into

mainfrom

Commits on Jul 5, 2026

Fix export-time trust_remote_code bypass in FP8/INT8/GGUF-LoRA export (#6869)

danielhanchen
and
pre-commit-ci[bot]
authored
64f6526

Commits on Jul 6, 2026

Harden Windows Pester install against missing PSGallery (#6892)
danielhanchen
authored
53a071c
Auto Xet to HTTP download fallback in from_pretrained; share Studio's fallback via unsloth_zoo (#6638)
danielhanchen
authored
cb6737c
GRPO: sequence packing for the no-grad old/ref logp path (default-on) (#6738)

authored
9407d49
Add PrefixGrouper for GRPO: dedup the shared prompt across a group's completions (#6871)

authored
08e133c
Handle odd shapes and non-float scales in FP8BlockQuantLinear (#6848)

danielhanchen
and
pre-commit-ci[bot]
authored
22bd86e
Scope MoE expert LoRA detection to actual MLP projection targets (#6849)

danielhanchen
and
pre-commit-ci[bot]
authored
7cc1752
Honor an explicit sdpa or flex_attention request when flash is disabled (#6847)

danielhanchen
and
pre-commit-ci[bot]
authored
c520662
Auto-enable grouped MoE on loaded / PEFT'd models via loader hook (#6727)

danielhanchen
and
pre-commit-ci[bot]
authored
c7b8666
Honor explicit load_in_16bit for local -bf16 directories (#6726)
danielhanchen
authored
95a73f0
Sync FORCE_FLOAT32 fallback with unsloth-zoo (gemma4, glm4_moe, qwen3_moe) (#6865)
danielhanchen
authored
efcaffb
Note bundled flash-linear-attention kernels for gated-deltanet models (#6850)
danielhanchen
authored
cf4906d
CI: pin lockfile-audit actions to commit SHAs (#6902)
anxkhn
authored
487b420
fix(fp8): use int64 offsets in weight_dequant_kernel (#6884)
anxkhn
authored
f4d1dc5
fix: map None quant method to q8_0 before lowercasing in GGUF export (#6889)
anxkhn
authored
c44d94f
fix: correct class name in SyntheticDataKit.chunk_data guard message (#6901)
anxkhn
authored
cc99aab
studio: label RAM and VRAM readouts as GiB not GB (#6895)
danielhanchen
authored
46e2cf5
Fix llama3 RoPE scaling dropped on transformers v5 (#6907)

danielhanchen
and
pre-commit-ci[bot]
authored
2fada48
Add the second blank line before _fix_rope_inv_freq (#6910)
danielhanchen
authored
cb9d902
studio: tool calling + healing parity for Llama-3, Mistral, Gemma 4 on safetensors + MLX (#5620)

authored
f0a5c52
Studio: stop chat generation on the assistant-turn-end token (fixes Qwen3.5 loop) (#6804)

danielhanchen
and
pre-commit-ci[bot]
authored
f38672d
studio: deterministic backend tool-calling wiring test (#6836)

danielhanchen
and
pre-commit-ci[bot]
authored
e9f49c6
Studio: coerce tool_call arguments to dict before chat templating (fixes MLX tool follow-up error) (#6807)

danielhanchen
and
pre-commit-ci[bot]
authored
e9ea45b
Studio: Gemma tool-call streaming follow-ups + nested-XML escape fix (#6476) (#6611)

danielhanchen
and
pre-commit-ci[bot]
authored
eb1ef44
studio: tool calling for DeepSeek (R1/V3/V3.1), GLM 4.x, Kimi K2 on safetensors + MLX (#5624)

authored
c00c1e7

Commits on Jul 7, 2026