OneTrainer icon indicating copy to clipboard operation
OneTrainer copied to clipboard

[Bug]: Lora - Rank and layers of prior LoRA model must match configuration

Open djp3k05 opened this issue 10 months ago • 7 comments

What happened?

Loading a Lora in OT that was trained with Alpha 16 (does not matter the rank), while trying to sample, will output to console the following errors:

\OneTrainer\venv\lib\site-packages\transformers\modeling_utils.py:5055: FutureWarning: `_is_quantized_training_enabled` is going to be deprecated in transformers 4.39.0. Please use `model.hf_quantizer.is_trainable` instead
  warnings.warn(
`loss_type=None` was set in the config but it is unrecognised.Using the default loss: `ForCausalLMLoss`.
Missing key for context_embedder.time_text_embed.timestep_embedder.linear_1; initializing it to zero.
Missing key for context_embedder.time_text_embed.timestep_embedder.linear_2; initializing it to zero.
Missing key for context_embedder.time_text_embed.text_embedder.linear_1; initializing it to zero.
Missing key for context_embedder.time_text_embed.text_embedder.linear_2; initializing it to zero.
Missing key for context_embedder.proj_in; initializing it to zero.
Missing key for context_embedder.token_refiner.refiner_blocks.0.attn.to_q; initializing it to zero.
Missing key for context_embedder.token_refiner.refiner_blocks.0.attn.to_k; initializing it to zero.
Missing key for context_embedder.token_refiner.refiner_blocks.0.attn.to_v; initializing it to zero.
Missing key for context_embedder.token_refiner.refiner_blocks.0.attn.to_out.0; initializing it to zero.
Missing key for context_embedder.token_refiner.refiner_blocks.0.ff.net.0.proj; initializing it to zero.
Missing key for context_embedder.token_refiner.refiner_blocks.0.ff.net.2; initializing it to zero.
Missing key for context_embedder.token_refiner.refiner_blocks.0.norm_out.linear; initializing it to zero.
Missing key for context_embedder.token_refiner.refiner_blocks.1.attn.to_q; initializing it to zero.
Missing key for context_embedder.token_refiner.refiner_blocks.1.attn.to_k; initializing it to zero.
Missing key for context_embedder.token_refiner.refiner_blocks.1.attn.to_v; initializing it to zero.
Missing key for context_embedder.token_refiner.refiner_blocks.1.attn.to_out.0; initializing it to zero.
Missing key for context_embedder.token_refiner.refiner_blocks.1.ff.net.0.proj; initializing it to zero.
Missing key for context_embedder.token_refiner.refiner_blocks.1.ff.net.2; initializing it to zero.
Missing key for context_embedder.token_refiner.refiner_blocks.1.norm_out.linear; initializing it to zero.
Missing key for time_text_embed.timestep_embedder.linear_1; initializing it to zero.
Missing key for time_text_embed.timestep_embedder.linear_2; initializing it to zero.
Missing key for time_text_embed.guidance_embedder.linear_1; initializing it to zero.
Missing key for time_text_embed.guidance_embedder.linear_2; initializing it to zero.
Missing key for time_text_embed.text_embedder.linear_1; initializing it to zero.
Missing key for time_text_embed.text_embedder.linear_2; initializing it to zero.
Missing key for transformer_blocks.0.norm1.linear; initializing it to zero.
Missing key for transformer_blocks.0.norm1_context.linear; initializing it to zero.
Missing key for transformer_blocks.0.attn.to_q; initializing it to zero.
Missing key for transformer_blocks.0.attn.to_k; initializing it to zero.
Missing key for transformer_blocks.0.attn.to_v; initializing it to zero.
Missing key for transformer_blocks.0.attn.add_k_proj; initializing it to zero.
Missing key for transformer_blocks.0.attn.add_v_proj; initializing it to zero.
Missing key for transformer_blocks.0.attn.add_q_proj; initializing it to zero.
Missing key for transformer_blocks.0.attn.to_out.0; initializing it to zero.
Missing key for transformer_blocks.0.attn.to_add_out; initializing it to zero.
Missing key for transformer_blocks.0.ff.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.0.ff.net.2; initializing it to zero.
Missing key for transformer_blocks.0.ff_context.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.0.ff_context.net.2; initializing it to zero.
Missing key for transformer_blocks.1.norm1.linear; initializing it to zero.
Missing key for transformer_blocks.1.norm1_context.linear; initializing it to zero.
Missing key for transformer_blocks.1.attn.to_q; initializing it to zero.
Missing key for transformer_blocks.1.attn.to_k; initializing it to zero.
Missing key for transformer_blocks.1.attn.to_v; initializing it to zero.
Missing key for transformer_blocks.1.attn.add_k_proj; initializing it to zero.
Missing key for transformer_blocks.1.attn.add_v_proj; initializing it to zero.
Missing key for transformer_blocks.1.attn.add_q_proj; initializing it to zero.
Missing key for transformer_blocks.1.attn.to_out.0; initializing it to zero.
Missing key for transformer_blocks.1.attn.to_add_out; initializing it to zero.
Missing key for transformer_blocks.1.ff.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.1.ff.net.2; initializing it to zero.
Missing key for transformer_blocks.1.ff_context.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.1.ff_context.net.2; initializing it to zero.
Missing key for transformer_blocks.2.norm1.linear; initializing it to zero.
Missing key for transformer_blocks.2.norm1_context.linear; initializing it to zero.
Missing key for transformer_blocks.2.attn.to_q; initializing it to zero.
Missing key for transformer_blocks.2.attn.to_k; initializing it to zero.
Missing key for transformer_blocks.2.attn.to_v; initializing it to zero.
Missing key for transformer_blocks.2.attn.add_k_proj; initializing it to zero.
Missing key for transformer_blocks.2.attn.add_v_proj; initializing it to zero.
Missing key for transformer_blocks.2.attn.add_q_proj; initializing it to zero.
Missing key for transformer_blocks.2.attn.to_out.0; initializing it to zero.
Missing key for transformer_blocks.2.attn.to_add_out; initializing it to zero.
Missing key for transformer_blocks.2.ff.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.2.ff.net.2; initializing it to zero.
Missing key for transformer_blocks.2.ff_context.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.2.ff_context.net.2; initializing it to zero.
Missing key for transformer_blocks.3.norm1.linear; initializing it to zero.
Missing key for transformer_blocks.3.norm1_context.linear; initializing it to zero.
Missing key for transformer_blocks.3.attn.to_q; initializing it to zero.
Missing key for transformer_blocks.3.attn.to_k; initializing it to zero.
Missing key for transformer_blocks.3.attn.to_v; initializing it to zero.
Missing key for transformer_blocks.3.attn.add_k_proj; initializing it to zero.
Missing key for transformer_blocks.3.attn.add_v_proj; initializing it to zero.
Missing key for transformer_blocks.3.attn.add_q_proj; initializing it to zero.
Missing key for transformer_blocks.3.attn.to_out.0; initializing it to zero.
Missing key for transformer_blocks.3.attn.to_add_out; initializing it to zero.
Missing key for transformer_blocks.3.ff.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.3.ff.net.2; initializing it to zero.
Missing key for transformer_blocks.3.ff_context.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.3.ff_context.net.2; initializing it to zero.
Missing key for transformer_blocks.4.norm1.linear; initializing it to zero.
Missing key for transformer_blocks.4.norm1_context.linear; initializing it to zero.
Missing key for transformer_blocks.4.attn.to_q; initializing it to zero.
Missing key for transformer_blocks.4.attn.to_k; initializing it to zero.
Missing key for transformer_blocks.4.attn.to_v; initializing it to zero.
Missing key for transformer_blocks.4.attn.add_k_proj; initializing it to zero.
Missing key for transformer_blocks.4.attn.add_v_proj; initializing it to zero.
Missing key for transformer_blocks.4.attn.add_q_proj; initializing it to zero.
Missing key for transformer_blocks.4.attn.to_out.0; initializing it to zero.
Missing key for transformer_blocks.4.attn.to_add_out; initializing it to zero.
Missing key for transformer_blocks.4.ff.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.4.ff.net.2; initializing it to zero.
Missing key for transformer_blocks.4.ff_context.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.4.ff_context.net.2; initializing it to zero.
Missing key for transformer_blocks.5.norm1.linear; initializing it to zero.
Missing key for transformer_blocks.5.norm1_context.linear; initializing it to zero.
Missing key for transformer_blocks.5.attn.to_q; initializing it to zero.
Missing key for transformer_blocks.5.attn.to_k; initializing it to zero.
Missing key for transformer_blocks.5.attn.to_v; initializing it to zero.
Missing key for transformer_blocks.5.attn.add_k_proj; initializing it to zero.
Missing key for transformer_blocks.5.attn.add_v_proj; initializing it to zero.
Missing key for transformer_blocks.5.attn.add_q_proj; initializing it to zero.
Missing key for transformer_blocks.5.attn.to_out.0; initializing it to zero.
Missing key for transformer_blocks.5.attn.to_add_out; initializing it to zero.
Missing key for transformer_blocks.5.ff.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.5.ff.net.2; initializing it to zero.
Missing key for transformer_blocks.5.ff_context.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.5.ff_context.net.2; initializing it to zero.
Missing key for transformer_blocks.6.norm1.linear; initializing it to zero.
Missing key for transformer_blocks.6.norm1_context.linear; initializing it to zero.
Missing key for transformer_blocks.6.attn.to_q; initializing it to zero.
Missing key for transformer_blocks.6.attn.to_k; initializing it to zero.
Missing key for transformer_blocks.6.attn.to_v; initializing it to zero.
Missing key for transformer_blocks.6.attn.add_k_proj; initializing it to zero.
Missing key for transformer_blocks.6.attn.add_v_proj; initializing it to zero.
Missing key for transformer_blocks.6.attn.add_q_proj; initializing it to zero.
Missing key for transformer_blocks.6.attn.to_out.0; initializing it to zero.
Missing key for transformer_blocks.6.attn.to_add_out; initializing it to zero.
Missing key for transformer_blocks.6.ff.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.6.ff.net.2; initializing it to zero.
Missing key for transformer_blocks.6.ff_context.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.6.ff_context.net.2; initializing it to zero.
Missing key for transformer_blocks.7.norm1.linear; initializing it to zero.
Missing key for transformer_blocks.7.norm1_context.linear; initializing it to zero.
Missing key for transformer_blocks.7.attn.to_q; initializing it to zero.
Missing key for transformer_blocks.7.attn.to_k; initializing it to zero.
Missing key for transformer_blocks.7.attn.to_v; initializing it to zero.
Missing key for transformer_blocks.7.attn.add_k_proj; initializing it to zero.
Missing key for transformer_blocks.7.attn.add_v_proj; initializing it to zero.
Missing key for transformer_blocks.7.attn.add_q_proj; initializing it to zero.
Missing key for transformer_blocks.7.attn.to_out.0; initializing it to zero.
Missing key for transformer_blocks.7.attn.to_add_out; initializing it to zero.
Missing key for transformer_blocks.7.ff.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.7.ff.net.2; initializing it to zero.
Missing key for transformer_blocks.7.ff_context.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.7.ff_context.net.2; initializing it to zero.
Missing key for transformer_blocks.8.norm1.linear; initializing it to zero.
Missing key for transformer_blocks.8.norm1_context.linear; initializing it to zero.
Missing key for transformer_blocks.8.attn.to_q; initializing it to zero.
Missing key for transformer_blocks.8.attn.to_k; initializing it to zero.
Missing key for transformer_blocks.8.attn.to_v; initializing it to zero.
Missing key for transformer_blocks.8.attn.add_k_proj; initializing it to zero.
Missing key for transformer_blocks.8.attn.add_v_proj; initializing it to zero.
Missing key for transformer_blocks.8.attn.add_q_proj; initializing it to zero.
Missing key for transformer_blocks.8.attn.to_out.0; initializing it to zero.
Missing key for transformer_blocks.8.attn.to_add_out; initializing it to zero.
Missing key for transformer_blocks.8.ff.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.8.ff.net.2; initializing it to zero.
Missing key for transformer_blocks.8.ff_context.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.8.ff_context.net.2; initializing it to zero.
Missing key for transformer_blocks.9.norm1.linear; initializing it to zero.
Missing key for transformer_blocks.9.norm1_context.linear; initializing it to zero.
Missing key for transformer_blocks.9.attn.to_q; initializing it to zero.
Missing key for transformer_blocks.9.attn.to_k; initializing it to zero.
Missing key for transformer_blocks.9.attn.to_v; initializing it to zero.
Missing key for transformer_blocks.9.attn.add_k_proj; initializing it to zero.
Missing key for transformer_blocks.9.attn.add_v_proj; initializing it to zero.
Missing key for transformer_blocks.9.attn.add_q_proj; initializing it to zero.
Missing key for transformer_blocks.9.attn.to_out.0; initializing it to zero.
Missing key for transformer_blocks.9.attn.to_add_out; initializing it to zero.
Missing key for transformer_blocks.9.ff.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.9.ff.net.2; initializing it to zero.
Missing key for transformer_blocks.9.ff_context.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.9.ff_context.net.2; initializing it to zero.
Missing key for transformer_blocks.10.norm1.linear; initializing it to zero.
Missing key for transformer_blocks.10.norm1_context.linear; initializing it to zero.
Missing key for transformer_blocks.10.attn.to_q; initializing it to zero.
Missing key for transformer_blocks.10.attn.to_k; initializing it to zero.
Missing key for transformer_blocks.10.attn.to_v; initializing it to zero.
Missing key for transformer_blocks.10.attn.add_k_proj; initializing it to zero.
Missing key for transformer_blocks.10.attn.add_v_proj; initializing it to zero.
Missing key for transformer_blocks.10.attn.add_q_proj; initializing it to zero.
Missing key for transformer_blocks.10.attn.to_out.0; initializing it to zero.
Missing key for transformer_blocks.10.attn.to_add_out; initializing it to zero.
Missing key for transformer_blocks.10.ff.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.10.ff.net.2; initializing it to zero.
Missing key for transformer_blocks.10.ff_context.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.10.ff_context.net.2; initializing it to zero.
Missing key for transformer_blocks.11.norm1.linear; initializing it to zero.
Missing key for transformer_blocks.11.norm1_context.linear; initializing it to zero.
Missing key for transformer_blocks.11.attn.to_q; initializing it to zero.
Missing key for transformer_blocks.11.attn.to_k; initializing it to zero.
Missing key for transformer_blocks.11.attn.to_v; initializing it to zero.
Missing key for transformer_blocks.11.attn.add_k_proj; initializing it to zero.
Missing key for transformer_blocks.11.attn.add_v_proj; initializing it to zero.
Missing key for transformer_blocks.11.attn.add_q_proj; initializing it to zero.
Missing key for transformer_blocks.11.attn.to_out.0; initializing it to zero.
Missing key for transformer_blocks.11.attn.to_add_out; initializing it to zero.
Missing key for transformer_blocks.11.ff.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.11.ff.net.2; initializing it to zero.
Missing key for transformer_blocks.11.ff_context.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.11.ff_context.net.2; initializing it to zero.
Missing key for transformer_blocks.12.norm1.linear; initializing it to zero.
Missing key for transformer_blocks.12.norm1_context.linear; initializing it to zero.
Missing key for transformer_blocks.12.attn.to_q; initializing it to zero.
Missing key for transformer_blocks.12.attn.to_k; initializing it to zero.
Missing key for transformer_blocks.12.attn.to_v; initializing it to zero.
Missing key for transformer_blocks.12.attn.add_k_proj; initializing it to zero.
Missing key for transformer_blocks.12.attn.add_v_proj; initializing it to zero.
Missing key for transformer_blocks.12.attn.add_q_proj; initializing it to zero.
Missing key for transformer_blocks.12.attn.to_out.0; initializing it to zero.
Missing key for transformer_blocks.12.attn.to_add_out; initializing it to zero.
Missing key for transformer_blocks.12.ff.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.12.ff.net.2; initializing it to zero.
Missing key for transformer_blocks.12.ff_context.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.12.ff_context.net.2; initializing it to zero.
Missing key for transformer_blocks.13.norm1.linear; initializing it to zero.
Missing key for transformer_blocks.13.norm1_context.linear; initializing it to zero.
Missing key for transformer_blocks.13.attn.to_q; initializing it to zero.
Missing key for transformer_blocks.13.attn.to_k; initializing it to zero.
Missing key for transformer_blocks.13.attn.to_v; initializing it to zero.
Missing key for transformer_blocks.13.attn.add_k_proj; initializing it to zero.
Missing key for transformer_blocks.13.attn.add_v_proj; initializing it to zero.
Missing key for transformer_blocks.13.attn.add_q_proj; initializing it to zero.
Missing key for transformer_blocks.13.attn.to_out.0; initializing it to zero.
Missing key for transformer_blocks.13.attn.to_add_out; initializing it to zero.
Missing key for transformer_blocks.13.ff.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.13.ff.net.2; initializing it to zero.
Missing key for transformer_blocks.13.ff_context.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.13.ff_context.net.2; initializing it to zero.
Missing key for transformer_blocks.14.norm1.linear; initializing it to zero.
Missing key for transformer_blocks.14.norm1_context.linear; initializing it to zero.
Missing key for transformer_blocks.14.attn.to_q; initializing it to zero.
Missing key for transformer_blocks.14.attn.to_k; initializing it to zero.
Missing key for transformer_blocks.14.attn.to_v; initializing it to zero.
Missing key for transformer_blocks.14.attn.add_k_proj; initializing it to zero.
Missing key for transformer_blocks.14.attn.add_v_proj; initializing it to zero.
Missing key for transformer_blocks.14.attn.add_q_proj; initializing it to zero.
Missing key for transformer_blocks.14.attn.to_out.0; initializing it to zero.
Missing key for transformer_blocks.14.attn.to_add_out; initializing it to zero.
Missing key for transformer_blocks.14.ff.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.14.ff.net.2; initializing it to zero.
Missing key for transformer_blocks.14.ff_context.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.14.ff_context.net.2; initializing it to zero.
Missing key for transformer_blocks.15.norm1.linear; initializing it to zero.
Missing key for transformer_blocks.15.norm1_context.linear; initializing it to zero.
Missing key for transformer_blocks.15.attn.to_q; initializing it to zero.
Missing key for transformer_blocks.15.attn.to_k; initializing it to zero.
Missing key for transformer_blocks.15.attn.to_v; initializing it to zero.
Missing key for transformer_blocks.15.attn.add_k_proj; initializing it to zero.
Missing key for transformer_blocks.15.attn.add_v_proj; initializing it to zero.
Missing key for transformer_blocks.15.attn.add_q_proj; initializing it to zero.
Missing key for transformer_blocks.15.attn.to_out.0; initializing it to zero.
Missing key for transformer_blocks.15.attn.to_add_out; initializing it to zero.
Missing key for transformer_blocks.15.ff.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.15.ff.net.2; initializing it to zero.
Missing key for transformer_blocks.15.ff_context.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.15.ff_context.net.2; initializing it to zero.
Missing key for transformer_blocks.16.norm1.linear; initializing it to zero.
Missing key for transformer_blocks.16.norm1_context.linear; initializing it to zero.
Missing key for transformer_blocks.16.attn.to_q; initializing it to zero.
Missing key for transformer_blocks.16.attn.to_k; initializing it to zero.
Missing key for transformer_blocks.16.attn.to_v; initializing it to zero.
Missing key for transformer_blocks.16.attn.add_k_proj; initializing it to zero.
Missing key for transformer_blocks.16.attn.add_v_proj; initializing it to zero.
Missing key for transformer_blocks.16.attn.add_q_proj; initializing it to zero.
Missing key for transformer_blocks.16.attn.to_out.0; initializing it to zero.
Missing key for transformer_blocks.16.attn.to_add_out; initializing it to zero.
Missing key for transformer_blocks.16.ff.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.16.ff.net.2; initializing it to zero.
Missing key for transformer_blocks.16.ff_context.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.16.ff_context.net.2; initializing it to zero.
Missing key for transformer_blocks.17.norm1.linear; initializing it to zero.
Missing key for transformer_blocks.17.norm1_context.linear; initializing it to zero.
Missing key for transformer_blocks.17.attn.to_q; initializing it to zero.
Missing key for transformer_blocks.17.attn.to_k; initializing it to zero.
Missing key for transformer_blocks.17.attn.to_v; initializing it to zero.
Missing key for transformer_blocks.17.attn.add_k_proj; initializing it to zero.
Missing key for transformer_blocks.17.attn.add_v_proj; initializing it to zero.
Missing key for transformer_blocks.17.attn.add_q_proj; initializing it to zero.
Missing key for transformer_blocks.17.attn.to_out.0; initializing it to zero.
Missing key for transformer_blocks.17.attn.to_add_out; initializing it to zero.
Missing key for transformer_blocks.17.ff.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.17.ff.net.2; initializing it to zero.
Missing key for transformer_blocks.17.ff_context.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.17.ff_context.net.2; initializing it to zero.
Missing key for transformer_blocks.18.norm1.linear; initializing it to zero.
Missing key for transformer_blocks.18.norm1_context.linear; initializing it to zero.
Missing key for transformer_blocks.18.attn.to_q; initializing it to zero.
Missing key for transformer_blocks.18.attn.to_k; initializing it to zero.
Missing key for transformer_blocks.18.attn.to_v; initializing it to zero.
Missing key for transformer_blocks.18.attn.add_k_proj; initializing it to zero.
Missing key for transformer_blocks.18.attn.add_v_proj; initializing it to zero.
Missing key for transformer_blocks.18.attn.add_q_proj; initializing it to zero.
Missing key for transformer_blocks.18.attn.to_out.0; initializing it to zero.
Missing key for transformer_blocks.18.attn.to_add_out; initializing it to zero.
Missing key for transformer_blocks.18.ff.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.18.ff.net.2; initializing it to zero.
Missing key for transformer_blocks.18.ff_context.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.18.ff_context.net.2; initializing it to zero.
Missing key for transformer_blocks.19.norm1.linear; initializing it to zero.
Missing key for transformer_blocks.19.norm1_context.linear; initializing it to zero.
Missing key for transformer_blocks.19.attn.to_q; initializing it to zero.
Missing key for transformer_blocks.19.attn.to_k; initializing it to zero.
Missing key for transformer_blocks.19.attn.to_v; initializing it to zero.
Missing key for transformer_blocks.19.attn.add_k_proj; initializing it to zero.
Missing key for transformer_blocks.19.attn.add_v_proj; initializing it to zero.
Missing key for transformer_blocks.19.attn.add_q_proj; initializing it to zero.
Missing key for transformer_blocks.19.attn.to_out.0; initializing it to zero.
Missing key for transformer_blocks.19.attn.to_add_out; initializing it to zero.
Missing key for transformer_blocks.19.ff.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.19.ff.net.2; initializing it to zero.
Missing key for transformer_blocks.19.ff_context.net.0.proj; initializing it to zero.
Missing key for transformer_blocks.19.ff_context.net.2; initializing it to zero.
Missing key for single_transformer_blocks.0.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.0.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.0.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.0.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.0.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.0.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.1.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.1.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.1.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.1.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.1.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.1.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.2.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.2.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.2.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.2.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.2.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.2.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.3.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.3.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.3.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.3.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.3.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.3.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.4.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.4.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.4.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.4.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.4.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.4.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.5.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.5.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.5.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.5.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.5.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.5.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.6.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.6.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.6.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.6.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.6.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.6.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.7.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.7.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.7.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.7.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.7.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.7.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.8.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.8.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.8.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.8.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.8.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.8.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.9.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.9.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.9.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.9.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.9.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.9.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.10.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.10.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.10.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.10.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.10.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.10.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.11.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.11.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.11.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.11.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.11.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.11.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.12.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.12.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.12.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.12.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.12.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.12.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.13.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.13.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.13.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.13.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.13.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.13.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.14.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.14.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.14.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.14.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.14.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.14.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.15.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.15.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.15.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.15.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.15.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.15.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.16.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.16.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.16.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.16.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.16.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.16.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.17.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.17.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.17.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.17.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.17.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.17.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.18.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.18.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.18.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.18.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.18.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.18.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.19.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.19.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.19.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.19.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.19.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.19.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.20.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.20.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.20.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.20.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.20.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.20.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.21.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.21.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.21.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.21.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.21.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.21.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.22.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.22.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.22.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.22.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.22.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.22.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.23.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.23.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.23.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.23.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.23.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.23.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.24.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.24.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.24.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.24.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.24.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.24.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.25.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.25.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.25.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.25.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.25.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.25.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.26.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.26.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.26.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.26.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.26.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.26.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.27.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.27.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.27.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.27.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.27.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.27.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.28.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.28.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.28.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.28.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.28.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.28.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.29.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.29.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.29.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.29.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.29.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.29.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.30.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.30.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.30.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.30.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.30.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.30.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.31.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.31.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.31.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.31.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.31.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.31.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.32.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.32.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.32.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.32.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.32.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.32.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.33.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.33.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.33.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.33.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.33.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.33.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.34.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.34.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.34.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.34.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.34.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.34.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.35.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.35.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.35.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.35.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.35.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.35.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.36.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.36.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.36.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.36.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.36.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.36.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.37.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.37.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.37.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.37.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.37.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.37.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.38.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.38.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.38.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.38.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.38.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.38.proj_out; initializing it to zero.
Missing key for single_transformer_blocks.39.attn.to_q; initializing it to zero.
Missing key for single_transformer_blocks.39.attn.to_k; initializing it to zero.
Missing key for single_transformer_blocks.39.attn.to_v; initializing it to zero.
Missing key for single_transformer_blocks.39.norm.linear; initializing it to zero.
Missing key for single_transformer_blocks.39.proj_mlp; initializing it to zero.
Missing key for single_transformer_blocks.39.proj_out; initializing it to zero.
Missing key for norm_out.linear; initializing it to zero.
Missing key for proj_out; initializing it to zero.
sampling: 100%|████████████████████████████████████████████████████████████████████████| 20/20 [00:25<00:00,  1.27s/it]

I used the default preset, only changed the optimizer to Adafactor. Tested numerous times: 16/16, 32/16, 64/16 will give that error. But any other combination like 16/1, 32/32, 64/32... works as expected. So my conclusion is that this is happening only when using Alpha 16. Sampling while training does works.

hunyan_def.json

What did you expect would happen?

Not getting errors.

Relevant log output


Output of pip freeze

No response

djp3k05 avatar Jan 27 '25 19:01 djp3k05

so... it seems that those settings for rank/alpha are not used only for training.... matching them with the Lora it fixes de errors. 😄 It is a bug or a feature?

djp3k05 avatar Jan 27 '25 21:01 djp3k05

so... it seems that those settings for rank/alpha are not used only for training.... matching them with the Lora it fixes de errors. 😄 It is a bug or a feature?

What do you mean matching them? You said they were already matched in training?

O-J1 avatar Feb 03 '25 03:02 O-J1

If a LoRA model file is loaded, the LoRA rank, layer filter, etc. are currently not loaded from the LoRA file. The LoRA file must match the configuration. @djp3k05 above has apparently change alpha and rank while still trying to load the same file. Alpha should not make a difference, but rank does.

Should we consider this a bug? enhancement? Is it even a useful enhancement, or would an improved error message alredy be sufficient?

dxqb avatar Feb 15 '25 14:02 dxqb

After several tests, not works good, i convert the model with alpha 16 and you will see the difference, with Lamine Yamal FC.Barcelona soccer Player trained with Onetrainer. Generated with SwarmUI, same parameters

Lora.py original not patched, model converted with alpha 16

https://github.com/user-attachments/assets/762b88ff-9f23-4b47-a524-38992b7e8800

Model onetrainer not converted, lora.py patched, not original

https://github.com/user-attachments/assets/1075d292-b910-47bd-9cef-69af552ca6ef

Original photo trained

Image

The first video it's similar to Lamine, the second it's Lamine

ivaxsirc avatar Feb 20 '25 09:02 ivaxsirc

After several tests, not works good, i convert the model with alpha 16 and you will see the difference, with Lamine Yamal FC.Barcelona soccer Player trained with Onetrainer. Generated with SwarmUI, same parameters

I don't see how this is related to the issue report above

dxqb avatar Feb 21 '25 14:02 dxqb

After several tests, not works good, i convert the model with alpha 16 and you will see the difference, with Lamine Yamal FC.Barcelona soccer Player trained with Onetrainer. Generated with SwarmUI, same parameters

I don't see how this is related to the issue report above

I understand that we are talking about Lora trainings for Hunyuan maded with Onetrainer

ivaxsirc avatar Feb 21 '25 14:02 ivaxsirc

Just for info, it also happens for Flux Lora, I didn't test with other models. Different rank/alpha, maybe layers too.

hyppyhyppo avatar Feb 28 '25 08:02 hyppyhyppo