qwen3.6-35b-a3b-miles-multievo-v2-iter034

Qwen3.6-35B-A3B MILES MultiEvo v2 — iteration 34

Reinforcement-learning checkpoint from the miles-multievo-v2 run on Qwen3.6-35B-A3B, taken at training iteration 34. Checkpoints are saved every 5 iterations; the published iterations of this run live together in one collection so you can compare points along the training curve.

Usage

import torch
from transformers import AutoModelForCausalLM, AutoProcessor

repo = "willamazon1/qwen3.6-35b-a3b-miles-multievo-v2-iter034"
proc = AutoProcessor.from_pretrained(repo)
model = AutoModelForCausalLM.from_pretrained(repo, dtype=torch.bfloat16, device_map="auto")

Conversion

Converted from a Megatron-LM torch_dist training checkpoint to HuggingFace safetensors with slime's tools/convert_torch_dist_to_hf.py, using --vocab-size 248320 to strip embedding padding and -a/--add-missing-from-origin-hf so the vision tower — which the language-model-only torch_dist checkpoint does not carry — is taken from the base model. MoE experts are stored in the grouped/fused layout (mlp.experts.gate_up_proj / down_proj).

Every shard was checked for NaN/Inf and the full tensor key set was diffed against a known-good conversion of this architecture before upload.