Quantization made by Richard Erkhov.

tinyllama-1.1b-sum-sft-full_v1.1 - GGUF

Model creator: https://huggingface.co/martimfasantos/
Original model: https://huggingface.co/martimfasantos/tinyllama-1.1b-sum-sft-full_v1.1/

Name	Quant method	Size
tinyllama-1.1b-sum-sft-full_v1.1.Q2_K.gguf	Q2_K	0.4GB
tinyllama-1.1b-sum-sft-full_v1.1.Q3_K_S.gguf	Q3_K_S	0.47GB
tinyllama-1.1b-sum-sft-full_v1.1.Q3_K.gguf	Q3_K	0.51GB
tinyllama-1.1b-sum-sft-full_v1.1.Q3_K_M.gguf	Q3_K_M	0.51GB
tinyllama-1.1b-sum-sft-full_v1.1.Q3_K_L.gguf	Q3_K_L	0.55GB
tinyllama-1.1b-sum-sft-full_v1.1.IQ4_XS.gguf	IQ4_XS	0.57GB
tinyllama-1.1b-sum-sft-full_v1.1.Q4_0.gguf	Q4_0	0.59GB
tinyllama-1.1b-sum-sft-full_v1.1.IQ4_NL.gguf	IQ4_NL	0.6GB
tinyllama-1.1b-sum-sft-full_v1.1.Q4_K_S.gguf	Q4_K_S	0.6GB
tinyllama-1.1b-sum-sft-full_v1.1.Q4_K.gguf	Q4_K	0.62GB
tinyllama-1.1b-sum-sft-full_v1.1.Q4_K_M.gguf	Q4_K_M	0.62GB
tinyllama-1.1b-sum-sft-full_v1.1.Q4_1.gguf	Q4_1	0.65GB
tinyllama-1.1b-sum-sft-full_v1.1.Q5_0.gguf	Q5_0	0.71GB
tinyllama-1.1b-sum-sft-full_v1.1.Q5_K_S.gguf	Q5_K_S	0.71GB
tinyllama-1.1b-sum-sft-full_v1.1.Q5_K.gguf	Q5_K	0.73GB
tinyllama-1.1b-sum-sft-full_v1.1.Q5_K_M.gguf	Q5_K_M	0.73GB
tinyllama-1.1b-sum-sft-full_v1.1.Q5_1.gguf	Q5_1	0.77GB
tinyllama-1.1b-sum-sft-full_v1.1.Q6_K.gguf	Q6_K	0.84GB
tinyllama-1.1b-sum-sft-full_v1.1.Q8_0.gguf	Q8_0	1.09GB

Original model description:

license: apache-2.0 base_model: TinyLlama/TinyLlama_v1.1 tags: - alignment-handbook - trl - sft - generated_from_trainer - trl - sft - generated_from_trainer datasets: - martimfasantos/openai-summarize-tldr model-index: - name: tinyllama-1.1b-sum-sft-full_v1.1 results: []

tinyllama-1.1b-sum-sft-full_v1.1

This model is a fine-tuned version of TinyLlama/TinyLlama_v1.1 on the martimfasantos/openai-summarize-tldr dataset. It achieves the following results on the evaluation set:

Loss: 2.1131

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: 8
eval_batch_size: 8
seed: 42
distributed_type: multi-GPU
gradient_accumulation_steps: 2
total_train_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: 1

Training results

Training Loss	Epoch	Step	Validation Loss
2.1116	0.9997	1476	2.1131

Framework versions

Transformers 4.41.2
Pytorch 2.1.2
Datasets 2.20.0
Tokenizers 0.19.1