Hugging Face's logo

NovoCode
/
Novocode7b-v3

Text Generation
Transformers
PyTorch
Safetensors
mistral
Generated from Trainer
text-generation-inference
Inference Endpoints
Model card Files Community
2
Novocode7b-v3 / README.md
leaderboard-pr-bot's picture
leaderboard-pr-bot
Adding Evaluation Results
513a7af verified
preview code
|
raw
history blame
4.38 kB
metadata
license: apache-2.0
tags:
  - generated_from_trainer
base_model: NovoCode/Novocode7b-v2
model-index:
  - name: out
    results: []

Built with Axolotl

See axolotl config

axolotl version: 0.4.0

base_model: NovoCode/Novocode7b-v2
model_type: MistralForCausalLM
tokenizer_type: LlamaTokenizer
is_mistral_derived_model: true

load_in_8bit: false
load_in_4bit: false
strict: false

datasets:
  - path: Intel/orca_dpo_pairs
    type:
      system_prompt: ""
      field_system: system
      field_instruction: question
      field_output: chosen
      field_output: rejected
      format: "[INST] {instruction} [/INST]"
      no_input_format: "[INST] {instruction} [/INST]"
dataset_prepared_path:
val_set_size: 0.05
output_dir: ./out

sequence_len: 8192
sample_packing: true
pad_to_sequence_len: true
eval_sample_packing: false

wandb_project:
wandb_entity:
wandb_watch:
wandb_name:
wandb_log_model:

gradient_accumulation_steps: 4
micro_batch_size: 2
num_epochs: 4
optimizer: adamw_bnb_8bit
lr_scheduler: cosine
learning_rate: 0.000005

train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: false

gradient_checkpointing: true
early_stopping_patience:
resume_from_checkpoint:
local_rank:
logging_steps: 1
xformers_attention:
flash_attention: true

warmup_steps: 10
evals_per_epoch: 4
eval_table_size:
eval_table_max_new_tokens: 128
saves_per_epoch: 1
debug:
deepspeed:
weight_decay: 0.0
fsdp:
fsdp_config:
special_tokens:
  bos_token: "<s>"
  eos_token: "</s>"
  unk_token: "<unk>"

out

This model is a fine-tuned version of NovoCode/Novocode7b-v2 on the Intel/orca_dpo_pairs dataset. It achieves the following results on the evaluation set:

  • Loss: 0.6792

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

  • learning_rate: 5e-06
  • train_batch_size: 2
  • eval_batch_size: 2
  • seed: 42
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 8
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: cosine
  • lr_scheduler_warmup_steps: 10
  • num_epochs: 4

Training results

Training Loss Epoch Step Validation Loss
0.7565 0.01 1 0.8244
0.4845 0.26 24 0.4685
0.4594 0.51 48 0.4435
0.4399 0.77 72 0.4284
0.3115 1.01 96 0.4221
0.2008 1.26 120 0.4614
0.2212 1.52 144 0.4552
0.2101 1.78 168 0.4516
0.119 2.02 192 0.4547
0.0925 2.27 216 0.5502
0.096 2.53 240 0.5751
0.0967 2.78 264 0.5774
0.0537 3.02 288 0.5765
0.0576 3.28 312 0.6687
0.0526 3.54 336 0.6786
0.0492 3.79 360 0.6792

Framework versions

  • Transformers 4.37.0
  • Pytorch 2.1.2+cu121
  • Datasets 2.16.1
  • Tokenizers 0.15.0

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric Value
Avg. 59.94
AI2 Reasoning Challenge (25-Shot) 57.51
HellaSwag (10-Shot) 81.17
MMLU (5-Shot) 61.91
TruthfulQA (0-shot) 48.29
Winogrande (5-shot) 74.51
GSM8k (5-shot) 36.24