[tml] Inkling model doc

[Zhichenzzz]

Adds the Inkling (975B) model recipe page under docs/models/thinkingmachines/, registers it in the models index and navigation. Companion to #1683.

[guapisolo]

LGTM.

[gemini-code-assist]

Code Review

This pull request introduces documentation for the Inkling model, a 975B multimodal MoE model from Thinking Machines. It updates the documentation index and navigation structure, and adds a comprehensive guide detailing Inkling's architecture, launch commands, and training recipes (including full-parameter, LoRA, and multimodal RL configurations). The review feedback suggests appending the user's username or UID to the temporary directories used for optimizer offloading to avoid permission conflicts on multi-user clusters.

Important

The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.

[gemini-code-assist]

When using shared temporary directories (such as /tmp) for caching or process-specific storage, it is recommended to append the user's username or UID (e.g., ${USER}) to the path. This prevents permission conflicts and directory sharing issues on multi-user clusters where multiple users might run the same launcher or script.

Suggested change

	--optimizer-state-nvme-dir /tmp/opt_offload
	--optimizer-state-nvme-chunk-mb 256
	--offload-train-target disk
	--offload-train-disk-dir /tmp/train_offload
	--optimizer-state-nvme-dir /tmp/opt_offload_${USER}
	--optimizer-state-nvme-chunk-mb 256
	--offload-train-target disk
	--offload-train-disk-dir /tmp/train_offload_${USER}

References

1. When using shared temporary directories (such as /tmp) for caching or process-specific storage, append the user's UID (e.g. using os.getuid() or environment variables like ${USER}) to the path to avoid permission conflicts and directory sharing issues on multi-user clusters.