Issues
is:issue state:open
is:issue state:open
Search results
[Bug] GLM-4.6v-Flash LoRA fine-tuning fails with NotImplementedError: get_input_embeddings
bugSomething isn't workingSomething isn't workingpendingThis problem is yet to be addressedThis problem is yet to be addressedStatus: Open.#9635 In hiyouga/LLaMA-Factory;自定义模型如何创建kt_optimize_rule
bugSomething isn't workingSomething isn't workingpendingThis problem is yet to be addressedThis problem is yet to be addressedStatus: Open.#9629 In hiyouga/LLaMA-Factory;Training hangs during backward pass with MoE models when some experts are not activated
bugSomething isn't workingSomething isn't workingpendingThis problem is yet to be addressedThis problem is yet to be addressedStatus: Open.#9628 In hiyouga/LLaMA-Factory;Ascend deepspeed zero3-offload 全参微调 Qwen3-VL-30B-A3B速度明显慢于 Qwen3-VL-32B模型
bugSomething isn't workingSomething isn't workingnpuThis problem is related to NPU devicesThis problem is related to NPU devicespendingThis problem is yet to be addressedThis problem is yet to be addressedStatus: Open.#9625 In hiyouga/LLaMA-Factory;PPO LoRA training with Qwen-14B on Ascend NPU: past_key_values NoneType error in generate (v0.9.4.dev0 + DeepSpeed)
bugSomething isn't workingSomething isn't workingnpuThis problem is related to NPU devicesThis problem is related to NPU devicespendingThis problem is yet to be addressedThis problem is yet to be addressedStatus: Open.#9623 In hiyouga/LLaMA-Factory;sft时 instruction 和 input是在哪个代码里拼起来的，没找到
bugSomething isn't workingSomething isn't workingpendingThis problem is yet to be addressedThis problem is yet to be addressedStatus: Open.#9614 In hiyouga/LLaMA-Factory;卡死不动， 多卡多机
bugSomething isn't workingSomething isn't workingpendingThis problem is yet to be addressedThis problem is yet to be addressedStatus: Open.#9603 In hiyouga/LLaMA-Factory;[RFC] Upgrade trl Dependency to Latest Version to Resolve Compatibility
enhancementNew feature or requestNew feature or requestpendingThis problem is yet to be addressedThis problem is yet to be addressedStatus: Open.#9601 In hiyouga/LLaMA-Factory;Invalid condition in "Dropped invalid example"
bugSomething isn't workingSomething isn't workingpendingThis problem is yet to be addressedThis problem is yet to be addressedStatus: Open.#9600 In hiyouga/LLaMA-Factory;lora 微调后，使用vllm_infer.py推理问题
bugSomething isn't workingSomething isn't workingpendingThis problem is yet to be addressedThis problem is yet to be addressedStatus: Open.#9598 In hiyouga/LLaMA-Factory;预训练Streaming 加载 fineweb 本地数据集
bugSomething isn't workingSomething isn't workingpendingThis problem is yet to be addressedThis problem is yet to be addressedStatus: Open.#9596 In hiyouga/LLaMA-Factory;RoPE scaling configuration not applied when using mcore_adapter for training
bugSomething isn't workingSomething isn't workingpendingThis problem is yet to be addressedThis problem is yet to be addressedStatus: Open.#9589 In hiyouga/LLaMA-Factory;