-
您好，使用原始代码在2张A100 80G上面微调qwen，显存占用两张卡上都只有919M，但是在数据加载过程中？内存占用一直在增加，直到180多G后内存爆了，程序终止。请问这个问题怎么解？
训练log：
:
- Is this model architecture supported by MLC-LLM? (the list of [supported models](https://llm.mlc.ai/do…
-
### Before Asking 在提问之前
- [X] I have read the [README](https://github.com/alibaba/data-juicer/blob/main/README.md) carefully. 我已经仔细阅读了 [README](https://github.com/alibaba/data-juicer/blob/main/README…
-
we have wanted to run qwen1.5 on our existing wormhole cards for some days. Happy to see Qwen1.5 appeared in the supported model list of the pybuda-0.19.3 just released，but it still only supports run…
-
### Required prerequisites
- [X] I have searched the [Issue Tracker](https://github.com/camel-ai/camel/issues) and [Discussions](https://github.com/camel-ai/camel/discussions) that this hasn't alre…
-
图片中标注的那部分能够换成千问
-
通义千问2-VL-2B-Instruct-GPTQ-Int4不支持多轮图片识别
错误提示：
{
"object": "error",
"message": "At most 1 image(s) may be provided in one request.",
"type": "BadRequestError",
"param": null,
…
-
How did you obtain the two model files, qwen-1.5-1.8b-chat-int8.mllm and qwen-1.5-1.8b-chat-q4k.mllm?
-
I see there is a Qwen-1_8B version of Instagger on ModelScope. Could you please share the prompt you used for finetuning this model so that we can obtain better results when using the tagger.
litsh updated
3 months ago
-
转换后的gguf模型, 无法处理qwen里面的如 等特殊标记, 分词的时候会出现把 分割为 "