- Rating
- 3
- Confidence
- 2
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
This work find that Long-context reasoning remains a critical bottleneck for large language models, as recent recurrent-memory approaches face two inherent challenges: sequential chunk-wise updates can overwrite early critical evidence with later irrelevant content, and serial inter-chunk dependencies limit parallelism and cause latency to increase with context length. Therefore, this ork propose PI-Mem that process all chunks ina parallel and iteratively refines a shared memory.
Critical strengths
- This work uses Figure 1 to explain the core idea, which is easy to understand.
- The paper propose the PI-memory, which use a shared memory so that the model could parallel process the long-context.
- The experiment support the effectiveness of PI-memory.
- The latency analysis proves that the PI-memory could be 16.9× faster than Vanilla and 6.1× faster than MemAgent, this suggests that the PI-memory could be deployed to speed up long context processing.
Critical weaknesses
- It seems that the model is Qwen, and it is not sure whether the method works well for other models.
- As the PI-memory is used to improve RL training for long-context. It is not sure whether the PI-memory is sensitive to the RL methods.
- What is the training cost of PI-memory, compared to baselines.
Questions for authors
NA
5 comments
Response to Reviewer iMWT (Part 2)
W2: Sensitivity to the RL method
To examine PI-Mem's sensitivity to the RL method, we additionally train Qwen2.5-7B PI-Mem with standard GRPO and Dr. GRPO configurations.
Because repeating full RL training is expensive, we conduct these additional experiments on Qwen2.5-7B for 80 steps, whereas the Qwen2.5 PI-Mem checkpoint reported in the paper was trained to step 240 (Appendix A, Table 5). All three configurations below are evaluated at step 80 on the eight HQA lengths from 7K to 896K, with 64 questions per length.
| Training configuration | Divide advantages by group std | Loss aggregation | KL coefficient | Reward: step 0 | Reward: step 80 | Reward gain | HQA 7K–896K (step 80) |
|---|---|---|---|---|---|---|---|
| Standard GRPO [1] | Yes | seq-mean-token-mean | 0.001 | 0.564 | 0.826 | +0.262 | 69.14 |
| Dr. GRPO [2] | No | seq-mean-token-sum-norm | No KL | 0.565 | 0.787 | +0.222 | 68.75 |
| Original configuration | No | token-mean | 0.001 | 0.551 | 0.799 | +0.248 | 69.34 |
All three configurations show substantial reward improvements and achieve similar downstream HQA scores, ranging from 68.75 to 69.34. This supports robustness across the tested RL configurations: the workflow can be trained effectively with both standard GRPO and Dr. GRPO.
Response to Reviewer iMWT (Part 3)
W3: Training cost and inference efficiency
Thank you for raising the concern about training cost. We acknowledge that PI-Mem requires more training time than MemAgent for the same number of training steps in our Qwen3.5 experiments.
PI-Mem takes longer to train in our Qwen3.5 experiments. We compare Qwen3.5-35B-A3B MemAgent and PI-Mem using the same training data and hyperparameters, as described in Appendix A. Both use 32 H200 GPUs and are trained for 80 steps. The training-time estimates from the logs are:
| Method | H200 GPUs | Training steps | Training time |
|---|---|---|---|
| MemAgent | 32 | 80 | 56 h 48 min 18 s |
| PI-Mem | 32 | 80 | 74 h 38 min 07 s |
PI-Mem takes approximately 31.4% longer. Its multi-turn workflow introduces additional processing during training, including repeated chunk reads and memory refinement. We acknowledge this additional training cost; these additional calls produce more call-level training sequences within each trajectory and increase the number of training tokens, while our efficiency claim concerns end-to-end inference latency, not faster training. For Qwen2.5 MemAgent, we use the officially released checkpoint and do not have a corresponding local training-time measurement. For context, a contributor to the official repository reports approximately 32 GPU-days for 150 training steps and continued performance improvements over the first 400 steps [3]. This comment does not specify the training step of the released checkpoint.
The additional training cost is distinct from inference latency. Appendix B.3, Table 11 reports the complete end-to-end inference workload, including all executed turns. For the 64-question HQA evaluation at 3.6M, using eight H200 GPUs with TP=2:
| Method | Qwen3.5 latency (s) | Qwen2.5 latency (s) |
|---|---|---|
| MemAgent | 12,060.71 | 5,535.50 |
| PI-Mem | 1,974.65 | 2,682.48 |
PI-Mem completes this workload faster by parallelizing chunk processing, despite performing more total FLOPs than MemAgent. The additional one-time training cost is therefore accompanied by lower latency during inference.
The workflow can also provide gains without additional training. The Gemma 4 E4B results above show a training-free HQA advantage over MemAgent. The existing Qwen3.5 results in Appendix B.2 show the same pattern: without RL, PI-Mem achieves 69.84 versus MemAgent's 64.37 on HQA (Table 7), and 96.25 versus 65.26 on RULER OOD tasks (Table 8). These results show that backbones with sufficient instruction-following and evidence-processing capabilities can already benefit from PI-Mem without additional training. RL remains useful for further improvement: Qwen3.5 PI-Mem reaches 81.25 on HQA and 98.04 on the OOD tasks after training.
References
[1] Zhihong Shao et al. DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models. 2024.
[2] Zichen Liu et al. Understanding R1-Zero-Like Training: A Critical Perspective. 2025.
[3] huiyeruzhou. Training-cost clarification in the official MemAgent repository, Issue #8. July 15, 2025.
Response to Reviewer iMWT (Part 1)
Dear reviewer iMWT,
Thank you for your careful review and constructive feedback. We address your concerns with additional experiments on a non-Qwen backbone, comparisons across RL training configurations, and a direct comparison of training costs.
W1: Generalization beyond Qwen
PI-Mem operates at the workflow level and is not tied to a specific LLM architecture, so it can be applied to different LLM backbones without architectural changes. Stronger instruction-following and evidence-processing capabilities help a model select relevant evidence and refine the shared memory, enabling it to make better use of PI-Mem for long-context tasks.
Training-free evaluation on Gemma 4 E4B. We apply Vanilla, MemAgent, and PI-Mem to the same instruction-tuned checkpoint
gemma-4-E4B-it, without any additional training. We evaluate 64 questions at each HQA length. Both memory workflows use 15K-token chunks and a 4,096-token output budget, with K=3 for PI-Mem. Vanilla uses the same output budget.Scores are Sub-EM (%). Length labels follow the paper's document-count settings; actual token counts depend on the tokenizer. PI-Mem achieves an average score of 65.00, compared with 58.75 for MemAgent. At the longest setting, PI-Mem scores 68.75 versus MemAgent's 50.00. These results demonstrate that the workflow's benefits extend beyond Qwen to Gemma 4 E4B without backbone-specific RL training.