Quantization made by Richard Erkhov.

Llama-2-7b-ultrachat-code - GGUF

Model creator: https://huggingface.co/kykim0/
Original model: https://huggingface.co/kykim0/Llama-2-7b-ultrachat-code/

Name	Quant method	Size
Llama-2-7b-ultrachat-code.Q2_K.gguf	Q2_K	2.36GB
Llama-2-7b-ultrachat-code.IQ3_XS.gguf	IQ3_XS	2.6GB
Llama-2-7b-ultrachat-code.IQ3_S.gguf	IQ3_S	2.75GB
Llama-2-7b-ultrachat-code.Q3_K_S.gguf	Q3_K_S	2.75GB
Llama-2-7b-ultrachat-code.IQ3_M.gguf	IQ3_M	2.9GB
Llama-2-7b-ultrachat-code.Q3_K.gguf	Q3_K	3.07GB
Llama-2-7b-ultrachat-code.Q3_K_M.gguf	Q3_K_M	3.07GB
Llama-2-7b-ultrachat-code.Q3_K_L.gguf	Q3_K_L	3.35GB
Llama-2-7b-ultrachat-code.IQ4_XS.gguf	IQ4_XS	3.4GB
Llama-2-7b-ultrachat-code.Q4_0.gguf	Q4_0	3.56GB
Llama-2-7b-ultrachat-code.IQ4_NL.gguf	IQ4_NL	3.58GB
Llama-2-7b-ultrachat-code.Q4_K_S.gguf	Q4_K_S	3.59GB
Llama-2-7b-ultrachat-code.Q4_K.gguf	Q4_K	3.8GB
Llama-2-7b-ultrachat-code.Q4_K_M.gguf	Q4_K_M	3.8GB
Llama-2-7b-ultrachat-code.Q4_1.gguf	Q4_1	3.95GB
Llama-2-7b-ultrachat-code.Q5_0.gguf	Q5_0	4.33GB
Llama-2-7b-ultrachat-code.Q5_K_S.gguf	Q5_K_S	4.33GB
Llama-2-7b-ultrachat-code.Q5_K.gguf	Q5_K	4.45GB
Llama-2-7b-ultrachat-code.Q5_K_M.gguf	Q5_K_M	4.45GB
Llama-2-7b-ultrachat-code.Q5_1.gguf	Q5_1	4.72GB
Llama-2-7b-ultrachat-code.Q6_K.gguf	Q6_K	5.15GB
Llama-2-7b-ultrachat-code.Q8_0.gguf	Q8_0	6.67GB

Original model description:

library_name: transformers base_model: kykim0/Llama-2-7b-ultrachat200k-2e tags: - alignment-handbook - trl - sft - generated_from_trainer - trl - sft - generated_from_trainer datasets: - HuggingFaceH4/ultrachat_200k model-index: - name: Llama-2-7b-ultrachat-code results: []

Llama-2-7b-ultrachat-code

This model is a fine-tuned version of kykim0/Llama-2-7b-ultrachat200k-2e on the HuggingFaceH4/ultrachat_200k dataset. It achieves the following results on the evaluation set:

Loss: 0.3676

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: 2e-05
train_batch_size: 1
eval_batch_size: 4
seed: 42
distributed_type: multi-GPU
num_devices: 4
gradient_accumulation_steps: 32
total_train_batch_size: 128
total_eval_batch_size: 16
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: cosine
lr_scheduler_warmup_ratio: 0.1
num_epochs: 2

Training results

Training Loss	Epoch	Step	Validation Loss
0.6476	0.9974	156	0.4590
0.4531	1.9948	312	0.3676

Framework versions

Transformers 4.44.2
Pytorch 2.1.2
Datasets 2.14.6
Tokenizers 0.19.1