---
tags: [agentic-ptb, grok]
base_model: Qwen/Qwen3.5-9B-Base
---

# grok.h072.sft-oh.step_40

AgentPTB sweep checkpoint. **Cell `grok` — pi / grok-4.6 @ effort `xhigh`.**

| field | value |
|---|---|
| plot cell | `grok` |
| driver | pi / grok-4.6 |
| reasoning effort | `xhigh` |
| run boot (UTC) | 2026-08-15T01:21:29Z |
| role | **intermediate** |
| **hours into run** | **h72.46** of 100 |
| checkpoint path in run | `outputs/sft-oh/weights/step_40` |
| shards | 4 |
| size | 18.8 GB |
| base model | `Qwen/Qwen3.5-9B-Base` |
| eos_token_id | `[248044, 248046]` ✅ correct (repaired 2026-08-28) |

## Reading the eos field

`248046` is `<|im_end|>`, the token the Qwen3.5 chat template ends every assistant turn with.
Checkpoints missing it do not stop at end-of-turn and overrun the context window, so their
eval numbers are a **floor, not a measurement** — compare them only against other checkpoints
with the same eos status, or re-package before evaluating.

**Cell note:** eos packaging defect across all checkpoints

## Mapping back to the figures

The repo id is `{cell}.h{HHH}.{family}.{step}`, where **`hHHH` is the hour of the 100-hour run
at which this checkpoint was written** — the *same x-axis* the sweep figures use for eval
panels (`t_h`). So a checkpoint drops onto the performance-over-time curve directly, and
sorting repo ids within a cell sorts them chronologically.

`hHHH` is rounded down to whole hours for sortability; the exact value is the
**hours into run** row above, and in `agentic-ptb/INDEX`.

## Packaging repair (2026-08-28)

`generation_config.json` now sets `eos_token_id` to `[248044, 248046]`, so generation stops on
both `<|endoftext|>` and the chat-format `<|im_end|>` token. Model weights are unchanged.