# If shard_idx=1+ confirm scheme matches prior shards.
            elif should_ignore_shard != should_ignore_layer:
                raise ValueError(f"Found a different quantization schemes for "
                                 f"{shard_proj_names} in {layer_name}. vLLM "
                                 "requires all to use the same scheme.")

from datasets import load_dataset
from loguru import logger
from transformers import AutoModelForCausalLM, AutoTokenizer

from llmcompressor.transformers import oneshot

# Select model and load it.
MODEL_ID = "microsoft/Phi-3-mini-4k-instruct"
model = AutoModelForCausalLM.from_pretrained(
    MODEL_ID,
    device_map="auto",
    torch_dtype="auto",
)
tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)

# Select calibration dataset.
DATASET_ID = "HuggingFaceH4/ultrachat_200k"
DATASET_SPLIT = "train_sft"

# Select number of samples. 512 samples is a good place to start.
# Increasing the number of samples can improve accuracy.
NUM_CALIBRATION_SAMPLES = 512
MAX_SEQUENCE_LENGTH = 2048

# Load dataset and preprocess.
ds = load_dataset(DATASET_ID, split=DATASET_SPLIT)
ds = ds.shuffle(seed=42).select(range(NUM_CALIBRATION_SAMPLES))


def process_and_tokenize(example):
    text = tokenizer.apply_chat_template(example["messages"], tokenize=False)
    return tokenizer(
        text,
        padding=False,
        max_length=MAX_SEQUENCE_LENGTH,
        truncation=True,
        add_special_tokens=False,
    )


ds = ds.map(process_and_tokenize, remove_columns=ds.column_names)

recipe = """
quant_stage:
  quant_modifiers:
    QuantizationModifier:
      ignore: ["lm_head"]
      kv_cache_scheme:
        {num_bits: 8, type: float, symmetric: true, strategy: tensor}
"""

# Apply algorithms.
oneshot(
    model=model,
    dataset=ds,
    recipe=recipe,
    max_seq_length=MAX_SEQUENCE_LENGTH,
    num_calibration_samples=NUM_CALIBRATION_SAMPLES,
)

# logger.info(
#     "Running sample generation. ",
#     "Note: Inference with the quantized kv_cache is not supported. ",
#     "Please use vLLM for inference with the quantized kv_cache.",
# )
# # Confirm generations of the quantized model look sane.
# print("\n\n")
# print("========== SAMPLE GENERATION ==============")
# input_ids = tokenizer("Hello my name is", return_tensors="pt").input_ids.to("cuda")
# output = model.generate(input_ids, max_new_tokens=100)
# print(tokenizer.decode(output[0]))
# print("==========================================\n\n")

# Save to disk compressed.
SAVE_DIR = MODEL_ID.split("/")[1] + "-FP8-KV-only"
model.save_pretrained(SAVE_DIR, save_compressed=True)
tokenizer.save_pretrained(SAVE_DIR)


from vllm import LLM
import torch
from vllm import LLM, SamplingParams
sampling_params = SamplingParams(temperature=0.80, top_p=0.95)

llm = LLM(model=SAVE_DIR)
outputs = llm.generate("Hello my name is", sampling_params)
print(outputs)

from vllm import LLM, SamplingParams

sampling_params = SamplingParams(temperature=0.80, top_p=0.95)

# https://huggingface.co/horheynm/Phi-3-mini-4k-instruct-kv_cache/blob/main/config.json
path = "horheynm/Phi-3-mini-4k-instruct-kv_cache"

llm = LLM(model=path)
outputs = llm.generate("Hello my name is", sampling_params)
print(outputs[0].outputs[0].text)