Upload 9 files

chat_template.jinja

@@ -0,0 +1,13 @@
+{{ bos_token }}
+{% if system %}<|System|>
+{{ system }}
+{% endif %}
+{% for message in messages %}
+{% if message.role == 'user' %}<|Benutzer|>
+{{ message.content }}
+{% elif message.role == 'assistant' %}<|Assistentin|>
+{{ message.content }}
+{% endif %}
+{% endfor %}
+{% if add_generation_prompt %}<|Assistentin|>
+{% else %}</s>{% endif %}

config.json

@@ -0,0 +1,81 @@
+{
+  "vocab_size": 35303,
+  "hidden_size": 1792,
+  "intermediate_size": 3584,
+  "num_hidden_layers": 18,
+  "num_attention_heads": 14,
+  "num_key_value_heads": 14,
+  "head_dim": 128,
+  "sliding_window": 4096,
+  "hidden_act": "silu",
+  "attention_dropout": 0.0,
+  "rms_norm_eps": 1e-05,
+  "initializer_range": 0.02,
+  "is_decoder": false,
+  "add_cross_attention": false,
+  "use_cache": false,
+  "attention_bias": true,
+  "num_local_experts": 1,
+  "num_experts_per_tok": 1,
+  "router_aux_loss_coef": 0.0,
+  "output_router_logits": false,
+  "layer_types": [
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 8192,
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "pad_token_id": 3,
+  "return_dict": true,
+  "output_hidden_states": false,
+  "dtype": "bfloat16",
+  "tie_word_embeddings": false,
+  "chunk_size_feed_forward": 0,
+  "is_encoder_decoder": false,
+  "cross_attention_hidden_size": null,
+  "tie_encoder_decoder": false,
+  "architectures": [
+    "EvaGptForCausalLM"
+  ],
+  "finetuning_task": null,
+  "id2label": {
+    "0": "LABEL_0",
+    "1": "LABEL_1"
+  },
+  "label2id": {
+    "LABEL_0": 0,
+    "LABEL_1": 1
+  },
+  "task_specific_params": null,
+  "problem_type": null,
+  "tokenizer_class": null,
+  "prefix": null,
+  "sep_token_id": null,
+  "decoder_start_token_id": null,
+  "_name_or_path": "./training_outputs/EvaGPT-German-0.7B_2026_7_2026-01-08_17-33-14_2026-01-08_19-10-49",
+  "transformers_version": "5.0.0.dev0",
+  "model_type": "eva_gpt",
+  "output_attentions": false,
+  "rope_scaling": {
+    "type": "yarn",
+    "factor": 2.0,
+    "original_max_position_embeddings": 4096
+  }
+}

generation_config.json

@@ -0,0 +1,7 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "pad_token_id": 3,
+  "transformers_version": "5.0.0.dev0"
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f27fd1030fcd576515fc7084bbdef8c97a855a685f12db604ebad0140f4bcfef
+size 1410052204

tokenizer.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dddf9d35fc8790d29daa471f4721338f6136df1b7543a05457425f005ff581fe
+size 753494

tokenizer_config.json

@@ -0,0 +1,18 @@
+{
+  "add_prefix_space": null,
+  "backend": "tokenizers",
+  "bos_token": "<s>",
+  "clean_up_tokenization_spaces": false,
+  "cls_token": "</s>",
+  "eos_token": "</s>",
+  "is_local": true,
+  "model_max_length": 1000000000000000019884624838656,
+  "model_specific_special_tokens": {},
+  "pad_token": "<pad>",
+  "sep_token": "</s>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "LlamaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}

train_config.json

@@ -0,0 +1,40 @@
+{
+  "model_dir": "./training_outputs/EvaGPT-German-0.7B_2026_7_2026-01-08_17-33-14",
+  "csv_path": "./datasets/output-100-3k.csv",
+  "device": "cuda",
+  "train_mode": "lora",
+  "learning_rate": 0.0002,
+  "lr_schedule": "cosine",
+  "per_device_train_batch_size": 1,
+  "gradient_accumulation_steps": 7,
+  "num_train_epochs": 1.0,
+  "max_steps": null,
+  "chunk_size": 4096,
+  "max_seq_length": 4096,
+  "template_mode": "dialogplus",
+  "column_name": "text",
+  "lora_r": 100,
+  "lora_alpha": 125,
+  "shuffle": true,
+  "sort_by_length": true,
+  "use_ngrams": false,
+  "ngram_max": 12,
+  "ngram_top_k": 1500,
+  "ngram_min_chars": 16,
+  "ngram_min_words": 2,
+  "ngram_max_samples": 4000,
+  "ngram_budgeted": true,
+  "ngram_target_fit": 0.98,
+  "ngram_eval_samples": 512,
+  "ngram_add_batch": 64,
+  "ngram_min_count": 2,
+  "ngram_max_token_chars": 384,
+  "ngram_max_tokens_per_text": 4096,
+  "precision_mode": "bf16",
+  "gradient_checkpointing": true,
+  "save_dir": "./training_outputs",
+  "merge_lora_on_save": true,
+  "dataloader_num_workers": 0,
+  "max_grad_norm": 1.0,
+  "weight_decay": 0.01
+}