PrunaAI

/

p1atdev-dart-v2-moe-sft-QUANTO-float8bit-smashed

like 0

Follow

Pruna AI 269

• "torch.FloatStorage",
• "transformers.models.mixtral.modeling_mixtral.MixtralSparseMoeBlock",
• "transformers.models.mixtral.modeling_mixtral.MixtralForCausalLM",
• "quanto.tensor.qtype.qtype",
• "torch.nn.modules.sparse.Embedding",
• "transformers.models.mixtral.modeling_mixtral.MixtralBlockSparseTop2MLP",
• "transformers.generation.configuration_utils.GenerationConfig",
• "torch._utils._rebuild_parameter",
• "transformers.models.mixtral.modeling_mixtral.MixtralSdpaAttention",
• "transformers.models.mixtral.modeling_mixtral.MixtralRotaryEmbedding",
• "torch.device",
• "torch.BFloat16Storage",
• "collections.OrderedDict",
• "transformers.models.mixtral.modeling_mixtral.MixtralRMSNorm",
• "torch.nn.modules.container.ModuleList",
• "transformers.models.mixtral.modeling_mixtral.MixtralModel",
• "quanto.nn.qlinear.QLinear",
• "torch.float8_e4m3fn",
• "torch._utils._rebuild_tensor_v2",
• "torch.nn.modules.activation.SiLU",
• "transformers.models.mixtral.configuration_mixtral.MixtralConfig",
• "torch.bfloat16",
• "__builtin__.set",
• "transformers.models.mixtral.modeling_mixtral.MixtralDecoderLayer"

How to fix it?