halcyon-llm

/

Llama-halcyon-1B-token-checkpoint-8000

like 0

Follow

Halcyon LLM 1

• "__builtin__.set",
• "torch.BFloat16Storage",
• "torch._utils._rebuild_tensor_v2",
• "collections.OrderedDict",
• "torch.Size"

How to fix it?

• "collections.OrderedDict",
• "deepspeed.utils.tensor_fragment.fragment_address",
• "deepspeed.runtime.zero.config.ZeroStageEnum",
• "torch.ByteStorage",
• "torch.FloatStorage",
• "torch._utils._rebuild_tensor_v2",
• "deepspeed.runtime.fp16.loss_scaler.LossScaler"

How to fix it?

• "torch.FloatStorage",
• "deepspeed.runtime.fp16.loss_scaler.LossScaler",
• "deepspeed.runtime.zero.config.ZeroStageEnum",
• "deepspeed.utils.tensor_fragment.fragment_address",
• "torch.ByteStorage",
• "torch._utils._rebuild_tensor_v2",
• "collections.OrderedDict"

How to fix it?

• "torch.FloatStorage",
• "deepspeed.runtime.fp16.loss_scaler.LossScaler",
• "deepspeed.runtime.zero.config.ZeroStageEnum",
• "deepspeed.utils.tensor_fragment.fragment_address",
• "torch.ByteStorage",
• "torch._utils._rebuild_tensor_v2",
• "collections.OrderedDict"

How to fix it?

• "torch.FloatStorage",
• "deepspeed.runtime.fp16.loss_scaler.LossScaler",
• "deepspeed.runtime.zero.config.ZeroStageEnum",
• "deepspeed.utils.tensor_fragment.fragment_address",
• "torch.ByteStorage",
• "torch._utils._rebuild_tensor_v2",
• "collections.OrderedDict"

How to fix it?