Views
No views yet
terminal-rl_qwen3-8b_8gpu_mixed-s6_asb2_ah2-rwnone_dense_dense-c0_dapo-ch0.28-tok1-dyn1_harness-camel-agent_mt10_2026-06-04_124602_0609copy/iter_0000343| field | value |
|---|---|
| base model | qwen3-8b |
| world size | 8 GPU |
| RL algo | mixed-s6, dapo (ch=0.28, tok=1, dyn=1) |
| reward | dense, dense-c0 |
| advantage | asb2, ah2-rwnone |
| harness | camel-agent |
| variant | mt10 |
.distcp files (~13.35 GiB each) + .metadata + common.pt + metadata.json.
Backend: torch_dist (sharded), torch (common). Total ~107 GiB.1__{rank}_{shard}.distcp for rank in 0..3, shard in 0..1
2.metadata
3common.pt
4metadata.json1import torch.distributed.checkpoint as dcp
2state_dict = {...} # template with empty tensors of expected shapes
3dcp.load(state_dict, checkpoint_id="path/to/iter_0000343")torch.save file:1python -m torch.distributed.checkpoint.format_utils \
2 dcp_to_torch_save path/to/iter_0000343 consolidated.pt