Lu Changqi
Lu Changqi
In different cpu topologies, run 'numactl -H | grep node [0-9]* cpus | sed node [0-9]* cpus: | head -{0} | cut-f1 -{1} -d.mat (self.num_numa_nodes, int(cores_per_node)) 'to get the id...
Hello,thank you for your code! when i am training ,there is no error .but can not train. CUDA devices: GPU numbers: 1 Create or load an imdb. Loaded dataset `voc_2007_trainval`...
When we perform fine tuning, benchmark is also very important, so we added a Python file called finetune_benchmark, which is specifically used for benchmark testing. In the test, we use...
When fine-tuning, performance indicators also need to be observed to measure the performance of related hardware, so relevant benchmark information is printed in train.
In vllm, when resources are insufficient, preemptive scheduling will be performed, so monitoring this indicator is also very important. New preempt related indicators are added to the metrics in RequestOutput,...
When benchmarking a LLM, token/s is also an important metric, so this metric is added to the llama benchmark.