Unable to Fine Tuning LLama2 for sequence classification

Beginners
1

I’ve been trying to train my own sequence classification model but get this error when trying to do it

I’ve been following this write up blog/Lora-for-sequence-classification-with-Roberta-Llama-Mistral.md at main · huggingface/blog · GitHub

RuntimeError: mat1 and mat2 shapes cannot be multiplied (416x4096 and 1x8388608)

can some one help me solve this.

compute_dtype = getattr(torch, “float16”)

bnb_config = BitsAndBytesConfig(
    load_in_4bit=True,
    bnb_4bit_quant_type="nf4",
    bnb_4bit_compute_dtype=compute_dtype,
    bnb_4bit_use_double_quant=False,
)

llama_model =  AutoModelForSequenceClassification.from_pretrained(
  pretrained_model_name_or_path=llama_checkpoint,
  num_labels=2,
  device_map= device, #"auto",
  offload_folder="offload",
  trust_remote_code=True,
  quantization_config=bnb_config
)

Llama config

LlamaConfig {
  "_name_or_path": "meta-llama/Llama-2-7b-hf",
  "architectures": [
    "LlamaForCausalLM"
  ],
  "attention_bias": false,
  "attention_dropout": 0.0,
  "bos_token_id": 1,
  "eos_token_id": 2,
  "hidden_act": "silu",
  "hidden_size": 4096,
  "initializer_range": 0.02,
  "intermediate_size": 11008,
  "max_position_embeddings": 4096,
  "model_type": "llama",
  "num_attention_heads": 32,
  "num_hidden_layers": 32,
  "num_key_value_heads": 32,
  "pad_token_id": 2,
  "pretraining_tp": 1,
  "quantization_config": {
    "_load_in_4bit": true,
    "_load_in_8bit": false,
    "bnb_4bit_compute_dtype": "float16",
    "bnb_4bit_quant_type": "nf4",
    "bnb_4bit_use_double_quant": false,
    "llm_int8_enable_fp32_cpu_offload": false,
    "llm_int8_has_fp16_weight": false,
    "llm_int8_skip_modules": null,
    "llm_int8_threshold": 6.0,
    "quant_method": "bitsandbytes"
  },
  "rms_norm_eps": 1e-05,
  "rope_scaling": null,
  "rope_theta": 10000.0,
  "tie_word_embeddings": false,
  "torch_dtype": "float16",
  "transformers_version": "4.38.0.dev0",
  "use_cache": false,
  "vocab_size": 32000
}