{ "accum_steps": 4, "auto_map": { "AutoModelForCausalLM": "modeling_fwkv.FWKVLanguageModel" }, "architectures": [ "FWKVLanguageModel" ], "batch_size": 8, "ce_chunk": 512, "d_emb": 128, "d_model": 512, "dataset_name": "HuggingFaceH4/ultrachat_200k", "dtype": "float32", "epochs": 2, "ffn_mult": 4, "grad_clip": 1.0, "lr": 0.0003, "max_train_examples": 20000, "max_val_examples": 1000, "model_type": "fwkv", "n_layers": 14, "seq_len": 1024, "tie_word_embeddings": true, "train_split": "train_sft", "transformers_version": "5.0.0", "val_split": "test_sft", "vocab_size": 50259, "wkv_floor": 0.1 }