|
About the 🤗Accelerate category
|
|
1
|
1671
|
February 20, 2022
|
|
How to train a >100GB model with hugging face trainer
|
|
0
|
9
|
March 28, 2023
|
|
No GPUs found in a machine definitely with GPUs
|
|
6
|
250
|
March 24, 2023
|
|
Accelerate test stuck on training
|
|
0
|
60
|
March 23, 2023
|
|
Log audio to comet_ml?
|
|
0
|
32
|
March 18, 2023
|
|
Good way to reshaffle/reacreate dataloader content?
|
|
0
|
33
|
March 18, 2023
|
|
Running inference on flan-ul2 on multi-gpu
|
|
7
|
253
|
March 17, 2023
|
|
How to save everything in one checkpoint?
|
|
2
|
85
|
March 17, 2023
|
|
NCCL Timeout Accelerate Load From Checkpoint
|
|
0
|
89
|
March 16, 2023
|
|
Meta device error while instantiating model
|
|
2
|
178
|
March 15, 2023
|
|
Infer_auto_device_map returns empty
|
|
2
|
600
|
March 15, 2023
|
|
Debugging accelerate processes on remote node(s)
|
|
0
|
65
|
March 13, 2023
|
|
How to only load model weights for the evalaution script?
|
|
1
|
83
|
March 13, 2023
|
|
Infrastructure for pretraining and finetuning via accelerate
|
|
0
|
47
|
March 13, 2023
|
|
Same number of optimizations steps with 1 GPU or 4 GPUs?
|
|
0
|
46
|
March 11, 2023
|
|
Question/Bug about accelerator.gather (how to use accelerate/accelerator.gather for contrastive learning)
|
|
1
|
95
|
March 9, 2023
|
|
Accelerator.backward(loss) never done!
|
|
3
|
73
|
March 9, 2023
|
|
Loading BloomForCausalLM from sharded checkpoints
|
|
7
|
175
|
March 8, 2023
|
|
Can't pickle error using accelerate multi-GPU
|
|
6
|
174
|
March 7, 2023
|
|
Replicating the same code in gpus
|
|
1
|
50
|
March 6, 2023
|
|
Perform knowledge distillation using accelerate
|
|
0
|
52
|
March 5, 2023
|
|
Use `accelerate` in SLURM environment
|
|
9
|
1025
|
March 3, 2023
|
|
No GPUs found in distributed mode
|
|
0
|
65
|
March 1, 2023
|
|
Weights & Biases sweep with multi gpu accelerate launch
|
|
3
|
699
|
February 28, 2023
|
|
Command died with <Signals.SIGSEGV: 11>
|
|
1
|
1211
|
February 28, 2023
|
|
Cannot create distributed environment
|
|
0
|
55
|
February 28, 2023
|
|
Constrain device map to GPUs
|
|
0
|
79
|
February 24, 2023
|
|
Bug with model.generate if max_length or max_new_tokens are set, with accelerate deepspeed zero level 3
|
|
3
|
119
|
February 21, 2023
|
|
Using gradient_accumulation_steps does not give the same results
|
|
0
|
162
|
February 18, 2023
|
|
Clarification on training metrics
|
|
0
|
89
|
February 10, 2023
|