WHATX
/

30k-Llama3-8B-Instruct

Model card Files Files and versions

30k-Llama3-8B-Instruct / checkpoint-40 /global_step40

729 MB

Ctrl+K

Ctrl+K

1 contributor

History: 1 commit

QJerry's picture

Upload entire folder 160 steps of 351 steps, half of 3 epochs.

aec27cf verified over 1 year ago

bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt
Detected Pickle imports (6)
- "collections.OrderedDict",
- "torch._utils._rebuild_tensor_v2",
- "deepspeed.runtime.zero.config.ZeroStageEnum",
- "deepspeed.runtime.fp16.loss_scaler.LossScaler",
- "torch.FloatStorage",
- "deepspeed.utils.tensor_fragment.fragment_address"
How to fix it?
529 MB
xet

Upload entire folder 160 steps of 351 steps, half of 3 epochs. over 1 year ago
mp_rank_00_model_states.pt
Detected Pickle imports (7)
- "torch.BFloat16Storage",
- "collections.OrderedDict",
- "torch._utils._rebuild_tensor_v2",
- "torch.Size",
- "torch.ByteStorage",
- "__builtin__.set",
- "torch.FloatStorage"
How to fix it?
200 MB
xet

Upload entire folder 160 steps of 351 steps, half of 3 epochs. over 1 year ago