Upload folder using huggingface_hub

README.md

@@ -1,3 +1,293 @@
----
-license: mit
----
+---
+language:
+  - ru
+library_name: transformers
+tags:
+  - text-generation
+  - gpt3
+  - russian
+  - causal-lm
+license: apache-2.0
+pipeline_tag: text-generation
+base_model: ai-forever/rugpt3xl
+---
+
+# ruGPT-3 XL (HuggingFace format)
+
+A 1.3B-parameter GPT-3-style language model for Russian, converted from the original
+[ai-forever/rugpt3xl](https://huggingface.co/ai-forever/rugpt3xl) Megatron-LM checkpoint
+into a native HuggingFace `transformers` format.
+
+This is a **base (pretrained) model**, not instruction-tuned. It performs text completion
+and can be fine-tuned for downstream tasks.
+
+## Model Details
+
+| Parameter | Value |
+|---|---|
+| Parameters | 1.3B |
+| Architecture | GPT-3 (decoder-only transformer) |
+| Hidden size | 2048 |
+| Layers | 24 |
+| Attention heads | 16 |
+| FFN intermediate size | 8192 |
+| Max sequence length | 2048 |
+| Vocabulary | 50,264 tokens (BPE) |
+| Activation | GELU |
+| Normalization | Pre-LayerNorm |
+| Position encoding | Learned absolute |
+| Precision | float16 |
+| Training data | 80B tokens of Russian text (4 epochs) |
+| Test perplexity | 12.05 |
+
+## Quick Start
+
+```python
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+model_name = "your-username/rugpt3xl-hf"  # replace with actual repo name
+
+tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
+model = AutoModelForCausalLM.from_pretrained(
+    model_name, trust_remote_code=True, device_map="auto"
+)
+
+inputs = tokenizer("Москва - столица", return_tensors="pt").to(model.device)
+outputs = model.generate(
+    **inputs,
+    max_new_tokens=100,
+    do_sample=True,
+    temperature=0.7,
+    top_p=0.9,
+    repetition_penalty=1.2,
+)
+print(tokenizer.decode(outputs[0], skip_special_tokens=True))
+```
+
+## Loading Options
+
+**GPU (float16, recommended):**
+
+```python
+model = AutoModelForCausalLM.from_pretrained(
+    model_name, trust_remote_code=True, device_map="auto"
+)
+```
+
+**CPU (float32):**
+
+```python
+import torch
+
+model = AutoModelForCausalLM.from_pretrained(
+    model_name, trust_remote_code=True, dtype=torch.float32, device_map="cpu"
+)
+```
+
+## Chat Template
+
+The tokenizer includes a simple chat template for question-answering:
+
+```python
+messages = [
+    {"role": "user", "content": "Какая столица России?"},
+]
+text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
+# Output: "Вопрос: Какая столица России?\n\nОтвет: "
+
+inputs = tokenizer(text, return_tensors="pt").to(model.device)
+outputs = model.generate(**inputs, max_new_tokens=100)
+print(tokenizer.decode(outputs[0], skip_special_tokens=True))
+```
+
+> **Note:** This is a base model, not an instruction-tuned chatbot. The chat template provides
+> a basic structure, but the model may not always follow instructions precisely. For reliable
+> conversational behavior, fine-tune the model on instruction/chat data.
+
+## Fine-tuning
+
+The model is fully compatible with standard HuggingFace training workflows.
+
+### Full Fine-tuning with Trainer
+
+```python
+from transformers import AutoModelForCausalLM, AutoTokenizer, Trainer, TrainingArguments
+
+model = AutoModelForCausalLM.from_pretrained(
+    model_name, trust_remote_code=True, device_map="auto"
+)
+tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
+
+args = TrainingArguments(
+    output_dir="./rugpt3xl-finetuned",
+    num_train_epochs=3,
+    per_device_train_batch_size=4,
+    gradient_accumulation_steps=4,
+    learning_rate=2e-5,
+    fp16=True,
+    save_strategy="epoch",
+    logging_steps=10,
+)
+
+trainer = Trainer(
+    model=model,
+    args=args,
+    train_dataset=your_dataset,  # dataset with input_ids, attention_mask, labels
+)
+trainer.train()
+```
+
+### LoRA Fine-tuning with PEFT
+
+```python
+from transformers import AutoModelForCausalLM, AutoTokenizer
+from peft import LoraConfig, get_peft_model, TaskType
+
+model = AutoModelForCausalLM.from_pretrained(
+    model_name, trust_remote_code=True, device_map="auto"
+)
+
+lora_config = LoraConfig(
+    task_type=TaskType.CAUSAL_LM,
+    r=16,
+    lora_alpha=32,
+    lora_dropout=0.05,
+    target_modules=["q_proj", "k_proj", "v_proj", "o_proj", "up_proj", "down_proj"],
+)
+model = get_peft_model(model, lora_config)
+model.print_trainable_parameters()
+# trainable params: ~14M || all params: 1.4B || trainable%: ~1.0%
+```
+
+### SFT with TRL
+
+```python
+from transformers import AutoModelForCausalLM, AutoTokenizer
+from trl import SFTTrainer, SFTConfig
+from peft import LoraConfig, TaskType
+from datasets import Dataset
+
+model = AutoModelForCausalLM.from_pretrained(
+    model_name, trust_remote_code=True, device_map="auto"
+)
+tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
+
+lora_config = LoraConfig(
+    task_type=TaskType.CAUSAL_LM,
+    r=16,
+    lora_alpha=32,
+    target_modules=["q_proj", "k_proj", "v_proj", "o_proj"],
+)
+
+# Dataset with chat messages format
+train_data = [
+    {"messages": [
+        {"role": "user", "content": "Какая столица России?"},
+        {"role": "assistant", "content": "Москва - столица Российской Федерации."},
+    ]},
+    # ... more examples
+]
+dataset = Dataset.from_list(train_data)
+
+sft_config = SFTConfig(
+    output_dir="./rugpt3xl-sft",
+    max_steps=1000,
+    per_device_train_batch_size=4,
+    learning_rate=2e-5,
+    logging_steps=10,
+    max_length=512,
+)
+
+trainer = SFTTrainer(
+    model=model,
+    args=sft_config,
+    train_dataset=dataset,
+    peft_config=lora_config,
+    processing_class=tokenizer,
+)
+trainer.train()
+```
+
+### Supported Fine-tuning Features
+
+| Feature | Status |
+|---|---|
+| Full parameter training | Supported |
+| Gradient checkpointing | Supported |
+| LoRA / PEFT | Supported |
+| TRL SFTTrainer | Supported |
+| DeepSpeed ZeRO | Supported |
+| FSDP | Supported |
+| KV cache during generation | Supported |
+| `labels` argument for loss computation | Supported |
+
+**LoRA target modules:** `q_proj`, `k_proj`, `v_proj`, `o_proj`, `up_proj`, `down_proj`
+
+## Architecture Details
+
+The model implements a custom `RuGPT3XLForCausalLM` class (loaded via `trust_remote_code=True`):
+
+```
+RuGPT3XLForCausalLM
+  ├── model (RuGPT3XLModel)
+  │     ├── embed_tokens      (Embedding: 50264 x 2048)
+  │     ├── embed_positions    (Embedding: 2048 x 2048)
+  │     ├── embed_dropout      (Dropout: 0.1)
+  │     ├── layers (x24)       (RuGPT3XLDecoderLayer)
+  │     │     ├── input_layernorm          (LayerNorm: 2048)
+  │     │     ├── self_attn                (RuGPT3XLAttention)
+  │     │     │     ├── q_proj             (Linear: 2048 -> 2048)
+  │     │     │     ├── k_proj             (Linear: 2048 -> 2048)
+  │     │     │     ├── v_proj             (Linear: 2048 -> 2048)
+  │     │     │     ├── o_proj             (Linear: 2048 -> 2048)
+  │     │     │     ├── attn_dropout       (Dropout: 0.1)
+  │     │     │     └── resid_dropout      (Dropout: 0.1)
+  │     │     ├── post_attention_layernorm  (LayerNorm: 2048)
+  │     │     └── mlp                      (RuGPT3XMLP)
+  │     │           ├── up_proj            (Linear: 2048 -> 8192)
+  │     │           ├── down_proj          (Linear: 8192 -> 2048)
+  │     │           ├── act_fn             (GELU)
+  │     │           └── dropout            (Dropout: 0.1)
+  │     └── norm               (LayerNorm: 2048)
+  └── lm_head                  (Linear: 2048 -> 50264, no bias)
+```
+
+## Conversion
+
+This model was converted from the original Megatron-LM checkpoint using a custom script.
+The conversion performs the following transformations:
+
+1. Strips the `module.` prefix from parameter names (FP16 / DDP wrappers)
+2. Remaps Megatron-LM naming to HuggingFace convention
+3. Splits the fused QKV projection (`[6144, 2048]`) into separate Q, K, V (`[2048, 2048]` each)
+4. Saves weights in safetensors format
+
+For full conversion details and the script, see the
+[rugpt3xl-convert](https://github.com/your-username/rugpt3xl-convert) repository.
+
+## Limitations
+
+- This is a **base model** trained on Russian internet text. It may generate biased, factually
+  incorrect, or offensive content.
+- The model was trained primarily on Russian text. It has limited capability in other languages.
+- Maximum context length is 2048 tokens. Inputs longer than this will be truncated.
+- The model is not instruction-tuned and works best for text completion rather than
+  following specific instructions.
+
+## Citation
+
+```bibtex
+@misc{rugpt3xl,
+  title={ruGPT-3 XL},
+  author={SberDevices Team},
+  year={2021},
+  publisher={Hugging Face},
+  url={https://huggingface.co/ai-forever/rugpt3xl}
+}
+```
+
+## Links
+
+- [ai-forever/rugpt3xl](https://huggingface.co/ai-forever/rugpt3xl) - original model
+- [ai-forever/ru-gpts](https://github.com/ai-forever/ru-gpts) - original training codebase
+- [GPT-3 Paper](https://arxiv.org/abs/2005.14165) - "Language Models are Few-Shot Learners" (Brown et al., 2020)

config.json

@@ -0,0 +1,30 @@
+{
+  "architectures": [
+    "RuGPT3XLForCausalLM"
+  ],
+  "auto_map": {
+    "AutoConfig": "configuration_rugpt3xl.RuGPT3XLConfig",
+    "AutoModel": "modeling_rugpt3xl.RuGPT3XLModel",
+    "AutoModelForCausalLM": "modeling_rugpt3xl.RuGPT3XLForCausalLM"
+  },
+  "model_type": "rugpt3xl",
+  "vocab_size": 50264,
+  "hidden_size": 2048,
+  "num_hidden_layers": 24,
+  "num_attention_heads": 16,
+  "intermediate_size": 8192,
+  "hidden_act": "gelu_new",
+  "max_position_embeddings": 2048,
+  "initializer_range": 0.02,
+  "layer_norm_eps": 1e-5,
+  "embedding_dropout": 0.1,
+  "attention_dropout": 0.1,
+  "output_dropout": 0.1,
+  "use_cache": true,
+  "bos_token_id": 2,
+  "eos_token_id": 1,
+  "pad_token_id": 0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float16",
+  "transformers_version": "5.3.0"
+}

configuration_rugpt3xl.py

@@ -0,0 +1,60 @@
+from transformers.configuration_utils import PretrainedConfig
+from transformers.utils import logging
+
+logger = logging.get_logger(__name__)
+
+
+class RuGPT3XLConfig(PretrainedConfig):
+    """Configuration class for the RuGPT-3 XL model (1.3B parameters).
+
+    This is a GPT-3-style decoder-only transformer trained on Russian text by
+    SberDevices. Architecture: learned absolute position embeddings, pre-norm
+    transformer layers with GELU activation, and tied word embeddings for the
+    language modeling head.
+    """
+
+    model_type = "rugpt3xl"
+    keys_to_ignore_at_inference = ["past_key_values"]
+
+    def __init__(
+        self,
+        vocab_size=50264,
+        hidden_size=2048,
+        num_hidden_layers=24,
+        num_attention_heads=16,
+        intermediate_size=8192,
+        hidden_act="gelu_new",
+        max_position_embeddings=2048,
+        initializer_range=0.02,
+        layer_norm_eps=1e-5,
+        embedding_dropout=0.1,
+        attention_dropout=0.1,
+        output_dropout=0.1,
+        use_cache=True,
+        bos_token_id=2,
+        eos_token_id=1,
+        pad_token_id=0,
+        tie_word_embeddings=False,
+        **kwargs,
+    ):
+        self.vocab_size = vocab_size
+        self.hidden_size = hidden_size
+        self.num_hidden_layers = num_hidden_layers
+        self.num_attention_heads = num_attention_heads
+        self.intermediate_size = intermediate_size
+        self.hidden_act = hidden_act
+        self.max_position_embeddings = max_position_embeddings
+        self.initializer_range = initializer_range
+        self.layer_norm_eps = layer_norm_eps
+        self.embedding_dropout = embedding_dropout
+        self.attention_dropout = attention_dropout
+        self.output_dropout = output_dropout
+        self.use_cache = use_cache
+
+        super().__init__(
+            bos_token_id=bos_token_id,
+            eos_token_id=eos_token_id,
+            pad_token_id=pad_token_id,
+            tie_word_embeddings=tie_word_embeddings,
+            **kwargs,
+        )

generation_config.json

@@ -0,0 +1,13 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 2,
+  "eos_token_id": 1,
+  "pad_token_id": 0,
+  "do_sample": true,
+  "temperature": 0.7,
+  "top_k": 50,
+  "top_p": 0.9,
+  "repetition_penalty": 1.2,
+  "max_new_tokens": 256,
+  "transformers_version": "5.3.0"
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:98a74e6d57c584e1192ff058be2efcab720c318481c4212e92d4d427abe52140
+size 2837399976

modeling_rugpt3xl.py

@@ -0,0 +1,503 @@
+"""PyTorch RuGPT-3 XL model.
+
+GPT-3-style decoder-only transformer (1.3B) trained on Russian text.
+Architecture: absolute position embeddings, pre-norm layers, GELU activation,
+tied LM head.
+"""
+
+import math
+from typing import List, Optional, Tuple, Union
+
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+import torch.utils.checkpoint
+
+from transformers.activations import ACT2FN
+from transformers.cache_utils import Cache, DynamicCache
+from transformers.modeling_outputs import (
+    BaseModelOutputWithPast,
+    CausalLMOutputWithPast,
+)
+from transformers.modeling_utils import PreTrainedModel
+from transformers.utils import logging
+
+from .configuration_rugpt3xl import RuGPT3XLConfig
+
+logger = logging.get_logger(__name__)
+
+
+class RuGPT3XLAttention(nn.Module):
+    def __init__(self, config: RuGPT3XLConfig, layer_idx: int):
+        super().__init__()
+        self.config = config
+        self.layer_idx = layer_idx
+        self.hidden_size = config.hidden_size
+        self.num_heads = config.num_attention_heads
+        self.head_dim = self.hidden_size // self.num_heads
+        self.scale = self.head_dim ** -0.5
+
+        self.q_proj = nn.Linear(self.hidden_size, self.hidden_size)
+        self.k_proj = nn.Linear(self.hidden_size, self.hidden_size)
+        self.v_proj = nn.Linear(self.hidden_size, self.hidden_size)
+        self.o_proj = nn.Linear(self.hidden_size, self.hidden_size)
+
+        self.attn_dropout = nn.Dropout(config.attention_dropout)
+        self.resid_dropout = nn.Dropout(config.output_dropout)
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_value: Optional[Cache] = None,
+        output_attentions: bool = False,
+        use_cache: bool = False,
+        **kwargs,
+    ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Cache]]:
+        bsz, q_len, _ = hidden_states.size()
+
+        query = self.q_proj(hidden_states)
+        key = self.k_proj(hidden_states)
+        value = self.v_proj(hidden_states)
+
+        query = query.view(bsz, q_len, self.num_heads, self.head_dim).transpose(1, 2)
+        key = key.view(bsz, q_len, self.num_heads, self.head_dim).transpose(1, 2)
+        value = value.view(bsz, q_len, self.num_heads, self.head_dim).transpose(1, 2)
+
+        if past_key_value is not None:
+            key, value = past_key_value.update(key, value, self.layer_idx)
+
+        attn_weights = torch.matmul(query, key.transpose(2, 3)) * self.scale
+
+        if attention_mask is not None:
+            attn_weights = attn_weights + attention_mask
+
+        attn_weights = F.softmax(attn_weights, dim=-1, dtype=torch.float32).to(
+            query.dtype
+        )
+        attn_weights = self.attn_dropout(attn_weights)
+
+        attn_output = torch.matmul(attn_weights, value)
+        attn_output = attn_output.transpose(1, 2).contiguous()
+        attn_output = attn_output.reshape(bsz, q_len, self.hidden_size)
+
+        attn_output = self.o_proj(attn_output)
+        attn_output = self.resid_dropout(attn_output)
+
+        return (
+            attn_output,
+            attn_weights if output_attentions else None,
+            past_key_value,
+        )
+
+
+class RuGPT3XMLP(nn.Module):
+    def __init__(self, config: RuGPT3XLConfig):
+        super().__init__()
+        self.up_proj = nn.Linear(config.hidden_size, config.intermediate_size)
+        self.down_proj = nn.Linear(config.intermediate_size, config.hidden_size)
+        self.act_fn = ACT2FN[config.hidden_act]
+        self.dropout = nn.Dropout(config.output_dropout)
+
+    def forward(self, hidden_states: torch.Tensor) -> torch.Tensor:
+        return self.dropout(self.down_proj(self.act_fn(self.up_proj(hidden_states))))
+
+
+class RuGPT3XLDecoderLayer(nn.Module):
+    def __init__(self, config: RuGPT3XLConfig, layer_idx: int):
+        super().__init__()
+        self.input_layernorm = nn.LayerNorm(
+            config.hidden_size, eps=config.layer_norm_eps
+        )
+        self.self_attn = RuGPT3XLAttention(config, layer_idx)
+        self.post_attention_layernorm = nn.LayerNorm(
+            config.hidden_size, eps=config.layer_norm_eps
+        )
+        self.mlp = RuGPT3XMLP(config)
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_value: Optional[Cache] = None,
+        output_attentions: bool = False,
+        use_cache: bool = False,
+        **kwargs,
+    ) -> Tuple[torch.Tensor, ...]:
+        # Pre-norm: LayerNorm -> Attention -> Residual
+        residual = hidden_states
+        hidden_states = self.input_layernorm(hidden_states)
+        hidden_states, self_attn_weights, present_key_value = self.self_attn(
+            hidden_states=hidden_states,
+            attention_mask=attention_mask,
+            position_ids=position_ids,
+            past_key_value=past_key_value,
+            output_attentions=output_attentions,
+            use_cache=use_cache,
+            **kwargs,
+        )
+        hidden_states = residual + hidden_states
+
+        # Pre-norm: LayerNorm -> MLP -> Residual
+        residual = hidden_states
+        hidden_states = self.post_attention_layernorm(hidden_states)
+        hidden_states = self.mlp(hidden_states)
+        hidden_states = residual + hidden_states
+
+        outputs = (hidden_states,)
+        if output_attentions:
+            outputs += (self_attn_weights,)
+        if use_cache:
+            outputs += (present_key_value,)
+        return outputs
+
+
+class RuGPT3XLPreTrainedModel(PreTrainedModel):
+    config_class = RuGPT3XLConfig
+    base_model_prefix = "model"
+    supports_gradient_checkpointing = True
+    _no_split_modules = ["RuGPT3XLDecoderLayer"]
+    _skip_keys_device_placement = ["past_key_values"]
+    _supports_cache_class = True
+
+    def _init_weights(self, module):
+        std = self.config.initializer_range
+        if isinstance(module, nn.Linear):
+            module.weight.data.normal_(mean=0.0, std=std)
+            if module.bias is not None:
+                module.bias.data.zero_()
+        elif isinstance(module, nn.Embedding):
+            module.weight.data.normal_(mean=0.0, std=std)
+            if module.padding_idx is not None:
+                module.weight.data[module.padding_idx].zero_()
+        elif isinstance(module, nn.LayerNorm):
+            module.bias.data.zero_()
+            module.weight.data.fill_(1.0)
+
+
+class RuGPT3XLModel(RuGPT3XLPreTrainedModel):
+    """Bare RuGPT-3 XL transformer outputting raw hidden states."""
+
+    def __init__(self, config: RuGPT3XLConfig):
+        super().__init__(config)
+        self.padding_idx = config.pad_token_id
+        self.vocab_size = config.vocab_size
+
+        self.embed_tokens = nn.Embedding(
+            config.vocab_size, config.hidden_size, self.padding_idx
+        )
+        self.embed_positions = nn.Embedding(
+            config.max_position_embeddings, config.hidden_size
+        )
+        self.embed_dropout = nn.Dropout(config.embedding_dropout)
+
+        self.layers = nn.ModuleList(
+            [
+                RuGPT3XLDecoderLayer(config, layer_idx)
+                for layer_idx in range(config.num_hidden_layers)
+            ]
+        )
+        self.norm = nn.LayerNorm(config.hidden_size, eps=config.layer_norm_eps)
+
+        self.gradient_checkpointing = False
+        self.post_init()
+
+    def get_input_embeddings(self):
+        return self.embed_tokens
+
+    def set_input_embeddings(self, value):
+        self.embed_tokens = value
+
+    def forward(
+        self,
+        input_ids: Optional[torch.LongTensor] = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[Union[Cache, List[torch.FloatTensor]]] = None,
+        inputs_embeds: Optional[torch.FloatTensor] = None,
+        use_cache: Optional[bool] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+        **kwargs,
+    ) -> Union[Tuple, BaseModelOutputWithPast]:
+        output_attentions = (
+            output_attentions
+            if output_attentions is not None
+            else self.config.output_attentions
+        )
+        output_hidden_states = (
+            output_hidden_states
+            if output_hidden_states is not None
+            else self.config.output_hidden_states
+        )
+        use_cache = use_cache if use_cache is not None else self.config.use_cache
+        return_dict = (
+            return_dict if return_dict is not None else self.config.use_return_dict
+        )
+
+        if input_ids is not None and inputs_embeds is not None:
+            raise ValueError(
+                "You cannot specify both input_ids and inputs_embeds at the same time"
+            )
+        if input_ids is not None:
+            batch_size, seq_length = input_ids.shape[:2]
+        elif inputs_embeds is not None:
+            batch_size, seq_length = inputs_embeds.shape[:2]
+        else:
+            raise ValueError("You have to specify either input_ids or inputs_embeds")
+
+        if self.gradient_checkpointing and self.training and use_cache:
+            logger.warning_once(
+                "`use_cache=True` is incompatible with gradient checkpointing. "
+                "Setting `use_cache=False`."
+            )
+            use_cache = False
+
+        past_key_values_length = 0
+        if use_cache:
+            if past_key_values is None:
+                past_key_values = DynamicCache()
+            past_key_values_length = past_key_values.get_seq_length()
+
+        if position_ids is None:
+            device = input_ids.device if input_ids is not None else inputs_embeds.device
+            position_ids = torch.arange(
+                past_key_values_length,
+                seq_length + past_key_values_length,
+                dtype=torch.long,
+                device=device,
+            )
+            position_ids = position_ids.unsqueeze(0)
+
+        if inputs_embeds is None:
+            inputs_embeds = self.embed_tokens(input_ids)
+
+        position_embeds = self.embed_positions(position_ids)
+        hidden_states = self.embed_dropout(inputs_embeds + position_embeds)
+
+        # Build causal 4D attention mask
+        causal_mask = self._build_causal_mask(
+            batch_size,
+            seq_length,
+            past_key_values_length,
+            hidden_states.dtype,
+            hidden_states.device,
+            attention_mask,
+        )
+
+        all_hidden_states = () if output_hidden_states else None
+        all_self_attns = () if output_attentions else None
+        next_decoder_cache = None
+
+        for decoder_layer in self.layers:
+            if output_hidden_states:
+                all_hidden_states += (hidden_states,)
+
+            if self.gradient_checkpointing and self.training:
+                layer_outputs = self._gradient_checkpointing_func(
+                    decoder_layer.__call__,
+                    hidden_states,
+                    causal_mask,
+                    position_ids,
+                    past_key_values,
+                    output_attentions,
+                    use_cache,
+                )
+            else:
+                layer_outputs = decoder_layer(
+                    hidden_states,
+                    attention_mask=causal_mask,
+                    position_ids=position_ids,
+                    past_key_value=past_key_values,
+                    output_attentions=output_attentions,
+                    use_cache=use_cache,
+                )
+
+            hidden_states = layer_outputs[0]
+            if use_cache:
+                next_decoder_cache = layer_outputs[2 if output_attentions else 1]
+            if output_attentions:
+                all_self_attns += (layer_outputs[1],)
+
+        hidden_states = self.norm(hidden_states)
+
+        if output_hidden_states:
+            all_hidden_states += (hidden_states,)
+
+        next_cache = None
+        if use_cache:
+            next_cache = next_decoder_cache
+
+        if not return_dict:
+            return tuple(
+                v
+                for v in [hidden_states, next_cache, all_hidden_states, all_self_attns]
+                if v is not None
+            )
+        return BaseModelOutputWithPast(
+            last_hidden_state=hidden_states,
+            past_key_values=next_cache,
+            hidden_states=all_hidden_states,
+            attentions=all_self_attns,
+        )
+
+    @staticmethod
+    def _build_causal_mask(
+        batch_size: int,
+        seq_length: int,
+        past_length: int,
+        dtype: torch.dtype,
+        device: torch.device,
+        attention_mask: Optional[torch.Tensor] = None,
+    ) -> torch.Tensor:
+        total_length = past_length + seq_length
+        causal = torch.full(
+            (seq_length, total_length), torch.finfo(dtype).min, device=device
+        )
+        causal = causal.masked_fill(
+            torch.arange(total_length, device=device).unsqueeze(0)
+            <= torch.arange(past_length, past_length + seq_length, device=device).unsqueeze(1),
+            0.0,
+        )
+        causal = causal.unsqueeze(0).unsqueeze(0)
+
+        if attention_mask is not None:
+            pad_mask = (1 - attention_mask[:, None, None, :].to(dtype)) * torch.finfo(
+                dtype
+            ).min
+            causal = causal + pad_mask
+
+        return causal
+
+
+class RuGPT3XLForCausalLM(RuGPT3XLPreTrainedModel):
+    _tied_weights_keys = {"lm_head.weight": "model.embed_tokens.weight"}
+
+    def __init__(self, config: RuGPT3XLConfig):
+        super().__init__(config)
+        self.model = RuGPT3XLModel(config)
+        self.vocab_size = config.vocab_size
+        self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)
+        self.post_init()
+
+    def get_input_embeddings(self):
+        return self.model.embed_tokens
+
+    def set_input_embeddings(self, value):
+        self.model.embed_tokens = value
+
+    def get_output_embeddings(self):
+        return self.lm_head
+
+    def set_output_embeddings(self, new_embeddings):
+        self.lm_head = new_embeddings
+
+    def get_decoder(self):
+        return self.model
+
+    def set_decoder(self, decoder):
+        self.model = decoder
+
+    def forward(
+        self,
+        input_ids: Optional[torch.LongTensor] = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[Union[Cache, List[torch.FloatTensor]]] = None,
+        inputs_embeds: Optional[torch.FloatTensor] = None,
+        labels: Optional[torch.LongTensor] = None,
+        use_cache: Optional[bool] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+        **kwargs,
+    ) -> Union[Tuple, CausalLMOutputWithPast]:
+        output_attentions = (
+            output_attentions
+            if output_attentions is not None
+            else self.config.output_attentions
+        )
+        output_hidden_states = (
+            output_hidden_states
+            if output_hidden_states is not None
+            else self.config.output_hidden_states
+        )
+        return_dict = (
+            return_dict if return_dict is not None else self.config.use_return_dict
+        )
+
+        outputs = self.model(
+            input_ids=input_ids,
+            attention_mask=attention_mask,
+            position_ids=position_ids,
+            past_key_values=past_key_values,
+            inputs_embeds=inputs_embeds,
+            use_cache=use_cache,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+
+        hidden_states = outputs[0]
+        logits = self.lm_head(hidden_states).float()
+
+        loss = None
+        if labels is not None:
+            shift_logits = logits[..., :-1, :].contiguous()
+            shift_labels = labels[..., 1:].contiguous()
+            loss_fct = nn.CrossEntropyLoss()
+            shift_logits = shift_logits.view(-1, self.config.vocab_size)
+            shift_labels = shift_labels.view(-1).to(shift_logits.device)
+            loss = loss_fct(shift_logits, shift_labels)
+
+        if not return_dict:
+            output = (logits,) + outputs[1:]
+            return (loss,) + output if loss is not None else output
+
+        return CausalLMOutputWithPast(
+            loss=loss,
+            logits=logits,
+            past_key_values=outputs.past_key_values,
+            hidden_states=outputs.hidden_states,
+            attentions=outputs.attentions,
+        )
+
+    def prepare_inputs_for_generation(
+        self,
+        input_ids,
+        past_key_values=None,
+        attention_mask=None,
+        inputs_embeds=None,
+        **kwargs,
+    ):
+        if past_key_values is not None:
+            past_length = past_key_values.get_seq_length()
+            if attention_mask is not None and attention_mask.shape[1] > input_ids.shape[1]:
+                input_ids = input_ids[:, -(attention_mask.shape[1] - past_length):]
+            elif past_length < input_ids.shape[1]:
+                input_ids = input_ids[:, past_length:]
+
+        position_ids = kwargs.get("position_ids", None)
+        if attention_mask is not None and position_ids is None:
+            position_ids = attention_mask.long().cumsum(-1) - 1
+            position_ids.masked_fill_(attention_mask == 0, 1)
+        if position_ids is not None and past_key_values is not None:
+            position_ids = position_ids[:, -input_ids.shape[1]:]
+
+        if inputs_embeds is not None and past_key_values is None:
+            model_inputs = {"inputs_embeds": inputs_embeds}
+        else:
+            model_inputs = {"input_ids": input_ids}
+
+        model_inputs.update(
+            {
+                "position_ids": position_ids,
+                "past_key_values": past_key_values,
+                "use_cache": kwargs.get("use_cache"),
+                "attention_mask": attention_mask,
+            }
+        )
+        return model_inputs

tokenizer_config.json

@@ -0,0 +1,13 @@
+{
+  "add_bos_token": false,
+  "add_eos_token": false,
+  "add_prefix_space": false,
+  "bos_token": "<s>",
+  "eos_token": "<|endoftext|>",
+  "pad_token": "<pad>",
+  "unk_token": "<unk>",
+  "model_max_length": 2048,
+  "tokenizer_class": "GPT2Tokenizer",
+  "clean_up_tokenization_spaces": true,
+  "chat_template": "{% for message in messages %}{% if message['role'] == 'system' %}{{ message['content'] + '\n\n' }}{% elif message['role'] == 'user' %}{{ 'Вопрос: ' + message['content'] + '\n\nОтвет: ' }}{% elif message['role'] == 'assistant' %}{{ message['content'] + eos_token }}{% endif %}{% endfor %}{% if add_generation_prompt %}{% endif %}"
+}