Config Files, Model, Tokenizer and README added.

README.md

@@ -1,3 +1,34 @@
 ---
+language: fa
 license: apache-2.0
+tags: 
+- Farsi
 ---
+# Arnavāz (ارنواز)
+
+**Model Description:** Arnavaz/gpt-arnavaz-beta is gpt2 language model that is fine-tuned using [bolbolzaban/gpt2-persian](https://huggingface.co/bolbolzaban/gpt2-persian) pretrained model.
+[bolbolzaban/gpt2-persian](https://huggingface.co/bolbolzaban/gpt2-persian) has been trained similar to [gpt2-medium](https://huggingface.co/gpt2-medium) with differences in context size, tokenizer and language [(Read more)](https://medium.com/@khashei/a-not-so-dangerous-ai-in-the-persian-language-39172a641c84).
+- **Developed by:** [Rezā Latifi](https://rezalatifi.ir)
+- **Model Type:** Transformer-based language model
+- **Language:** Persian (All characters other than the Persian alphabet are replaced with special tokens)
+- **License:** [Modified MIT License](https://github.com/openai/gpt-2/blob/master/LICENSE)
+- **Related Models:** [bolbolzaban/gpt2-persian](https://huggingface.co/bolbolzaban/gpt2-persian), [gpt2-medium](https://huggingface.co/gpt2-medium)
+- **Resources for more information:**
+  - [Arnavaz Website](https://openai.com/blog/better-language-models/)
+
+## How to utilize
+Using a pipeline for text generation, Arnavaz can be utilized like this:
+
+```python
+from transformers import pipeline, AutoTokenizer, GPT2LMHeadModel, AutoConfig
+tokenizer = AutoTokenizer.from_pretrained('Arnavaz/gpt2-arnavaz-beta')
+model = GPT2LMHeadModel.from_pretrained('Arnavaz/gpt2-arnavaz-beta')
+config = AutoConfig.from_pretrained('Arnavaz/gpt2-arnavaz-beta', max_length=512)
+generator = pipeline('text-generation', model, tokenizer=tokenizer, config=config)
+
+def getEloquent(ineloquent):
+  result = generator(f"[BOS]{ineloquent}[SEP]")[0]['generated_text']
+  return result[result.find('[SEP]')+5:]
+ 
+sample = getEloquent('استفاده از کاغذ پاپیروس برای نوشتن کتاب از حدود دو هزار سال قبل از میلاد در مصر رایج شد.')
+```

config.json

@@ -0,0 +1,65 @@
+{
+  "_name_or_path": "./",
+  "activation_function": "gelu_new",
+  "architectures": [
+    "GPT2LMHeadModel"
+  ],
+  "attn_dropout": 0.1,
+  "attn_pdrop": 0.1,
+  "beta1": 0.9,
+  "beta2": 0.98,
+  "bos_token_id": 8,
+  "data_path": "",
+  "embd_pdrop": 0.1,
+  "embed_dropout": 0.1,
+  "eos_token_id": 9,
+  "epsilon": 1e-09,
+  "eval_batch_size": 128,
+  "eval_steps": 10,
+  "gradient_checkpointing": false,
+  "initializer_range": 0.02,
+  "iterations": 1000,
+  "layer_norm_epsilon": 1e-05,
+  "line_count": 1,
+  "lr": 0.00025,
+  "max_steps": 3600000,
+  "model": "GPT2",
+  "model_type": "gpt2",
+  "n_ctx": 256,
+  "n_embd": 1024,
+  "n_head": 16,
+  "n_inner": null,
+  "n_layer": 24,
+  "n_positions": 256,
+  "n_vocab": 25000,
+  "opt_name": "adam",
+  "precision": "float32",
+  "predict_batch_size": 8,
+  "reorder_and_upcast_attn": false,
+  "res_dropout": 0.1,
+  "resid_pdrop": 0.1,
+  "scale_attn_by_inverse_layer_idx": false,
+  "scale_attn_weights": true,
+  "scale_by_depth": true,
+  "scale_by_in": true,
+  "summary_activation": null,
+  "summary_first_dropout": 0.1,
+  "summary_proj_to_labels": true,
+  "summary_type": "cls_index",
+  "summary_use_proj": true,
+  "task_specific_params": {
+    "text-generation": {
+      "do_sample": true,
+      "max_length": 512
+    }
+  },
+  "tokenizer_class": "AlbertTokenizer",
+  "torch_dtype": "float32",
+  "train_batch_size": 320,
+  "train_steps": 10000,
+  "transformers_version": "4.12.3",
+  "use_cache": true,
+  "vocab_size": 25000,
+  "warmup_steps": 2000,
+  "weight_decay": 0.01
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1ff44b8480d9c7d23509f2d395457e77d224313d2c2a7ce6420b2a86c018953e
+size 1314389977

tokenizer_config.json

@@ -0,0 +1,17 @@
+{
+  "do_lower_case": false,
+  "remove_space": true,
+  "keep_accents": true,
+  "bos_token": "[BOS]",
+  "eos_token": "[EOS]",
+  "unk_token": "[UNK]",
+  "sep_token": "[SEP]",
+  "pad_token": "[PAD]",
+  "cls_token": "[CLS]",
+  "mask_token": "[MSK]",
+  "normalizer": {
+    "type": "Sequence",
+    "normalizers": []
+  },
+  "name_or_path": "./"
+}