fidelity-root
glm53-flash-fidelity-root-v1
fidelity--glm53flash.malaiwah.root.bf16
A root fidelity dataset in hidden form, produced by engines/tools/hf_capture.py from zai-org/GLM-5.3-Flash-BF16.
The cut
the final hidden state handed to lm_head -- after the text model's final norm and immediately before the head matmul -- captured as the head module's input via torch.nn.Module.register_forward_pre_hook; replay applies the head ONLY (no final norm at replay time: the capture already sits after it). Same cut… See the full description on the dataset page: https://huggingface.co/datasets/malaiwah/glm53-flash-fidelity-root-v1.glm53-fidelity-root-v1
fidelity--glm53.malaiwah.root.bf16
A root fidelity dataset in hidden form, produced by engines/tools/hf_capture.py from zai-org/GLM-5.3-BF16.
The cut
the final hidden state handed to lm_head -- after the text model's final norm and immediately before the head matmul -- captured as the head module's input via torch.nn.Module.register_forward_pre_hook; replay applies the head ONLY (no final norm at replay time: the capture already sits after it). Same cut as… See the full description on the dataset page: https://huggingface.co/datasets/malaiwah/glm53-fidelity-root-v1.qwen38-27b-fidelity-root-v1
Qwen3.8-27B BF16 root fidelity dataset (hidden form)
A root capture of Qwen/Qwen3.8-27B at revision
1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0 — 18 shards, no quantization_config,
a genuinely unquantized reference — over the sealed suite-v5 shard-0 token
panel (512 contexts x 2048 tokens = 1,048,064 scored positions).
What this is for
Quantization fidelity is usually reported as a KL divergence against a teacher.
If the teacher was captured on a different stack than… See the full description on the dataset page: https://huggingface.co/datasets/malaiwah/qwen38-27b-fidelity-root-v1.glm52-fidelity-root-v1
fidelity--glm52.malaiwah.root.bf16
A root fidelity dataset in hidden form, produced by engines/tools/hf_capture.py from zai-org/GLM-5.2.
The cut
the final hidden state handed to lm_head -- after the text model's final norm and immediately before the head matmul -- captured as the head module's input via torch.nn.Module.register_forward_pre_hook; replay applies the head ONLY (no final norm at replay time: the capture already sits after it). Same cut as… See the full description on the dataset page: https://huggingface.co/datasets/malaiwah/glm52-fidelity-root-v1.qwen3-5-tiny-fidelity-root-v1
qwen3_5 random CPU fixture root
A root fidelity dataset in hidden form, produced by engines/tools/hf_capture.py from malaiwah/qwen3-5-tiny-random-bf16.
The cut
the final hidden state handed to lm_head -- after the text model's final norm and immediately before the head matmul -- captured as the head module's input via torch.nn.Module.register_forward_pre_hook; replay applies the head ONLY (no final norm at replay time: the capture already sits after it). Same cut… See the full description on the dataset page: https://huggingface.co/datasets/malaiwah/qwen3-5-tiny-fidelity-root-v1.deepseek-v4-tiny-fidelity-root-v1
deepseek-v4 random CPU fixture root
A root fidelity dataset in hidden form, produced by engines/tools/hf_capture.py from malaiwah/deepseek-v4-tiny-random-bf16.
The cut
the final hidden state handed to lm_head -- after the text model's final norm and immediately before the head matmul -- captured as the head module's input via torch.nn.Module.register_forward_pre_hook; replay applies the head ONLY (no final norm at replay time: the capture already sits after it).… See the full description on the dataset page: https://huggingface.co/datasets/malaiwah/deepseek-v4-tiny-fidelity-root-v1.