Upload 3 files

_Train_Model.py

@@ -0,0 +1,196 @@
+from transformers import AutoTokenizer, AutoModelForCausalLM, TextGenerationPipeline, AutoConfig, BitsAndBytesConfig,AutoConfig
+import time
+import torch
+torch.backends.cuda.matmul.allow_tf32 = True
+import random
+from datasets import load_dataset
+from transformers import TrainingArguments
+from trl import SFTTrainer
+from peft import LoraConfig
+# from accelerate import infer_auto_device_map, init_empty_weights, dispatch_model
+from torch.nn import CrossEntropyLoss
+torch.autograd.set_detect_anomaly(True)
+random_seed = 42
+torch.manual_seed(random_seed)
+random.seed(random_seed)
+# Set the device for each process
+device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+# torch.cuda.set_device(device)
+
+
+
+n_ahead_talk_global = 4
+n_passes_global = 2
+n_ahead_global = 8
+n_examples = 0
+
+def model_init(params):
+    original = False
+    if params is None:
+        params = {}
+    else:
+        params = params.params
+    # save params to file
+    n_ahead = params.get("n_ahead", n_ahead_global if not original else 1)
+    n_ahead_talk = params.get("n_ahead_talk", n_ahead_talk_global if not original else 1)
+    n_passes = params.get("n_passes", n_passes_global if not original else 1)
+    gumbel_temperature = params.get("gumbel_temperature", 1)
+    use_start_thought_token = params.get("use_start_thought_token", True)
+    use_end_thought_token = params.get("use_end_thought_token", True)
+    include_policy_loss = params.get("include_policy_loss", True)
+    gumbel_detach = params.get("gumbel_detach", True)
+    merged_talk_heads = params.get("merged_talk_heads", True)
+    residual_think_head = params.get("residual_think_head", False)
+    optimize_lm_head_only_at_start = params.get("optimize_lm_head_only_at_start", False)
+
+    model_id = "LeroyDyer/SpydazWeb_AGI_MistralStar"
+    tokenizer_id = model_id
+    print("Loading model")
+
+    model = AutoModelForCausalLM.from_pretrained(
+        model_id,
+        torch_dtype=torch.bfloat16 if torch.cuda.is_available() else torch.float32,
+        max_thoughts=n_ahead + n_ahead_talk + 1,
+        merged_talk_heads=merged_talk_heads,
+        merged_lm_and_talk_heads=False,
+        merged_lm_and_think_heads=True,
+        use_concat_talk_head=True,
+        use_shallow_think=True,
+        use_shallow_talk=False,
+        use_complex_think_head=False,
+        use_complex_talk_head=True,
+        use_weighted_talk_head=True,
+        trust_remote_code=True,
+        device_map="auto",
+    )
+    print("Loaded model")
+
+    tokenizer = AutoTokenizer.from_pretrained(tokenizer_id, truncation=True, padding_side="right")
+    tokenizer.pad_token_id = tokenizer.eos_token_id
+
+
+
+    model.gumbel_detach = gumbel_detach
+    model.include_policy_loss = include_policy_loss
+    model.use_end_thought_token = use_end_thought_token
+    model.use_start_thought_token = use_start_thought_token
+    model.n_ahead = n_ahead
+    model.n_ahead_talk = n_ahead_talk
+    model.n_passes = n_passes
+    model.residual_think_head = residual_think_head
+    model.optimize_lm_head_only_at_start = optimize_lm_head_only_at_start
+    model.gumbel_temperature = gumbel_temperature
+    model.original_mode = original
+    model.config_params = params
+    return model
+    
+model,tokenizer = model_init(None)
+
+
+## TRAINING : 
+
+peft_config = LoraConfig(
+          r = 128, # Choose any number > 0 ! Suggested 8, 16, 32, 64, 128
+    target_modules = ["q_proj", "k_proj", "v_proj", "o_proj",
+                      "gate_proj", "up_proj", "down_proj","lm_head", "embed_tokens"],
+    lora_alpha = 32,
+    lora_dropout = 0, # Supports any, but = 0 is optimized
+    bias = "none", 
+    use_dora=True,
+)
+
+from transformers import AutoTokenizer, AutoModelForCausalLM, TextGenerationPipeline, AutoConfig
+from datasets import load_dataset
+from transformers import TrainingArguments
+from trl import SFTTrainer
+from peft import LoraConfig
+
+## DATA
+alpaca_prompt = """Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request.
+
+### Instruction:
+{}
+
+### Input:
+{}
+
+### Response:
+{}"""
+EOS_TOKEN = tokenizer.eos_token # Must add EOS_TOKEN
+def formatting_prompts_func(examples):
+    instructions = examples["instruction"]
+    inputs       = examples["input"]
+    outputs      = examples["output"]
+    texts = []
+    for instruction, input, output in zip(instructions, inputs, outputs):
+        # Must add EOS_TOKEN, otherwise your generation will go on forever!
+        text = alpaca_prompt.format(instruction, input, output) + EOS_TOKEN
+        texts.append(text)
+    return { "text" : texts, }
+pass
+dataset = load_dataset("gate369/Alpaca-Star", split = "train[:2000]")
+dataset = dataset.shuffle(seed=3704)
+dataset = dataset.map(formatting_prompts_func, batched = True,)
+## TRAIN
+
+max_seq_length = 1024
+training_args = TrainingArguments(
+    output_dir="./out",
+    num_train_epochs=3,
+    per_device_train_batch_size=1,
+    gradient_checkpointing=False,
+    gradient_accumulation_steps=8,
+    optim="lion_32bit",
+    logging_steps=1,
+    save_strategy="steps",
+    max_steps=1000,
+    bf16=True,
+    tf32=False,
+    learning_rate=6e-05,
+    max_grad_norm=0.3,
+    warmup_ratio=0.06,
+    lr_scheduler_type="cosine",
+    push_to_hub=False,
+
+)
+trainer = SFTTrainer(
+    args=training_args,
+    train_dataset=dataset,
+    model=model,
+    tokenizer=tokenizer,
+    max_seq_length=max_seq_length,
+    dataset_text_field="text",
+    peft_config=peft_config,
+)
+trainer.train()
+
+## SAVE
+tokenizer.save_pretrained("SFTTrainerModel")
+model.save_pretrained("SFTTrainerModel")
+
+
+import os
+import huggingface_hub
+from huggingface_hub import notebook_login
+from huggingface_hub import create_repo, HfApi
+from huggingface_hub import hf_hub_download
+from huggingface_hub import create_repo, HfApi
+from huggingface_hub import snapshot_download
+
+MODEL_NAME = "_Spydaz_Web_AI_MistralStar"
+Folderinput = "SFTTrainerModel"
+WRITE_TOKEN = ""
+username = "LeroyDyer"
+huggingface_hub.login(WRITE_TOKEN)
+api = HfApi(token=WRITE_TOKEN)
+# Create empty repo
+api.create_repo(
+    repo_id = f"{username}/{MODEL_NAME}",
+    repo_type="model",
+    exist_ok=True,
+)
+
+api.upload_folder(
+    repo_id = f"{username}/{MODEL_NAME}",
+    folder_path = Folderinput
+)

configuration_mistral.py

@@ -12,7 +12,7 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
-""" Mistral model configuration"""
+"""Mistral model configuration"""
 
 from ...configuration_utils import PretrainedConfig
 from ...utils import logging
@@ -20,11 +20,6 @@ from ...utils import logging
 
 logger = logging.get_logger(__name__)
 
-MISTRAL_PRETRAINED_CONFIG_ARCHIVE_MAP = {
-    "mistralai/Mistral-7B-v0.1": "https://huggingface.co/mistralai/Mistral-7B-v0.1/resolve/main/config.json",
-    "mistralai/Mistral-7B-Instruct-v0.1": "https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.1/resolve/main/config.json",
-}
-
 
 class MistralConfig(PretrainedConfig):
     r"""
@@ -163,6 +158,152 @@ class MistralConfig(PretrainedConfig):
         self.use_complex_talk_head = use_complex_talk_head
         self.use_weighted_talk_head = use_weighted_talk_head
 
+        super().__init__(
+            pad_token_id=pad_token_id,
+            bos_token_id=bos_token_id,
+            eos_token_id=eos_token_id,
+            tie_word_embeddings=tie_word_embeddings,
+            **kwargs,
+        )
+class MistralStarConfig(PretrainedConfig):
+    r"""
+    This is the configuration class to store the configuration of a [`MistralModel`]. It is used to instantiate an
+    Mistral model according to the specified arguments, defining the model architecture. Instantiating a configuration
+    with the defaults will yield a similar configuration to that of the Mistral-7B-v0.1 or Mistral-7B-Instruct-v0.1.
+
+    [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1)
+    [mistralai/Mistral-7B-Instruct-v0.1](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.1)
+
+    Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
+    documentation from [`PretrainedConfig`] for more information.
+
+
+    Args:
+        vocab_size (`int`, *optional*, defaults to 32000):
+            Vocabulary size of the Mistral model. Defines the number of different tokens that can be represented by the
+            `inputs_ids` passed when calling [`MistralModel`]
+        hidden_size (`int`, *optional*, defaults to 4096):
+            Dimension of the hidden representations.
+        intermediate_size (`int`, *optional*, defaults to 14336):
+            Dimension of the MLP representations.
+        num_hidden_layers (`int`, *optional*, defaults to 32):
+            Number of hidden layers in the Transformer encoder.
+        num_attention_heads (`int`, *optional*, defaults to 32):
+            Number of attention heads for each attention layer in the Transformer encoder.
+        num_key_value_heads (`int`, *optional*, defaults to 8):
+            This is the number of key_value heads that should be used to implement Grouped Query Attention. If
+            `num_key_value_heads=num_attention_heads`, the model will use Multi Head Attention (MHA), if
+            `num_key_value_heads=1 the model will use Multi Query Attention (MQA) otherwise GQA is used. When
+            converting a multi-head checkpoint to a GQA checkpoint, each group key and value head should be constructed
+            by meanpooling all the original heads within that group. For more details checkout [this
+            paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to `8`.
+        hidden_act (`str` or `function`, *optional*, defaults to `"silu"`):
+            The non-linear activation function (function or string) in the decoder.
+        max_position_embeddings (`int`, *optional*, defaults to `4096*32`):
+            The maximum sequence length that this model might ever be used with. Mistral's sliding window attention
+            allows sequence of up to 4096*32 tokens.
+        initializer_range (`float`, *optional*, defaults to 0.02):
+            The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
+        rms_norm_eps (`float`, *optional*, defaults to 1e-06):
+            The epsilon used by the rms normalization layers.
+        use_cache (`bool`, *optional*, defaults to `True`):
+            Whether or not the model should return the last key/values attentions (not used by all models). Only
+            relevant if `config.is_decoder=True`.
+        pad_token_id (`int`, *optional*):
+            The id of the padding token.
+        bos_token_id (`int`, *optional*, defaults to 1):
+            The id of the "beginning-of-sequence" token.
+        eos_token_id (`int`, *optional*, defaults to 2):
+            The id of the "end-of-sequence" token.
+        tie_word_embeddings (`bool`, *optional*, defaults to `False`):
+            Whether the model's input and output word embeddings should be tied.
+        rope_theta (`float`, *optional*, defaults to 10000.0):
+            The base period of the RoPE embeddings.
+        sliding_window (`int`, *optional*, defaults to 4096):
+            Sliding window attention window size. If not specified, will default to `4096`.
+        attention_dropout (`float`, *optional*, defaults to 0.0):
+            The dropout ratio for the attention probabilities.
+
+    ```python
+    >>> from transformers import MistralModel, MistralConfig
+
+    >>> # Initializing a Mistral 7B style configuration
+    >>> configuration = MistralConfig()
+
+    >>> # Initializing a model from the Mistral 7B style configuration
+    >>> model = MistralModel(configuration)
+
+    >>> # Accessing the model configuration
+    >>> configuration = model.config
+    ```"""
+
+    model_type = "mistralstar"
+    keys_to_ignore_at_inference = ["past_key_values"]
+
+    def __init__(
+        self,
+        vocab_size=32000,
+        hidden_size=4096,
+        intermediate_size=14336,
+        num_hidden_layers=32,
+        num_attention_heads=32,
+        num_key_value_heads=8,
+        hidden_act="silu",
+        max_position_embeddings=4096 * 32,
+        initializer_range=0.02,
+        rms_norm_eps=1e-6,
+        use_cache=True,
+        pad_token_id=None,
+        bos_token_id=1,
+        eos_token_id=2,
+        tie_word_embeddings=False,
+        rope_theta=10000.0,
+        sliding_window=4096,
+        attention_dropout=0.0,
+        max_thoughts=16,
+        thought_length = 10,
+        merged_talk_heads=True,
+        merged_lm_and_talk_heads=False,
+        merged_lm_and_think_heads=True,
+        use_concat_talk_head=True,
+        use_shallow_think=True,
+        use_shallow_talk=False,
+        use_complex_think_head=False,
+        use_complex_talk_head=True,
+        use_weighted_talk_head=True,
+        **kwargs,
+    ):
+        self.vocab_size = vocab_size
+        self.max_position_embeddings = max_position_embeddings
+        self.hidden_size = hidden_size
+        self.intermediate_size = intermediate_size
+        self.num_hidden_layers = num_hidden_layers
+        self.num_attention_heads = num_attention_heads
+        self.sliding_window = sliding_window
+
+        # for backward compatibility
+        if num_key_value_heads is None:
+            num_key_value_heads = num_attention_heads
+
+        self.num_key_value_heads = num_key_value_heads
+        self.hidden_act = hidden_act
+        self.initializer_range = initializer_range
+        self.rms_norm_eps = rms_norm_eps
+        self.use_cache = use_cache
+        self.rope_theta = rope_theta
+        self.attention_dropout = attention_dropout
+        self.max_thoughts = max_thoughts
+        self.thought_length = thought_length
+        self.merged_talk_heads = merged_talk_heads
+        self.merged_lm_and_talk_heads = merged_lm_and_talk_heads
+        self.merged_lm_and_think_heads = merged_lm_and_think_heads
+        self.use_concat_talk_head = use_concat_talk_head
+        self.use_shallow_think = use_shallow_think
+        self.use_shallow_talk = use_shallow_talk
+        self.use_complex_think_head = use_complex_think_head
+        self.use_complex_talk_head = use_complex_talk_head
+        self.use_weighted_talk_head = use_weighted_talk_head
+
         super().__init__(
             pad_token_id=pad_token_id,
             bos_token_id=bos_token_id,