Downloads · 30 days
69
2% of all-time downloads
stablelm-2-brief-1_6b
maxim-saplin/stablelm-2-brief-1_6b
stablelm-2-brief-1_6b is a text generation model from maxim-saplin. Use it when you need the model to write or continue text. It is set up for transformers. The card lists the license as other.
Fine-tuned chat model based on stabilityai/stablelm-2-16b. It was trained for 21 epochs using 1024 context windows and a mixture of small subsets of UltraChat and OASST2. Since the model saw shorter dialogs it tends t…
All-time downloads
2.9K
Public
Parameters
1.6B
8.3 GB on disk
Likes
1
Public
Repo makeup
Click a slice to open those files.
.gguf5 GB · 60%
At a glance
- Task
- Text Generation
- Library
- transformers
- License
- other
- Model type
- stablelm
- Access
- Public
- Created
- Apr 13, 2024
- Updated
- Apr 16, 2024
- SHA
- 97b29e3b
Try a prompt
Model card
From the Hugging Face model README
Stable LM 2 BRIEF 1.6B
Fine-tuned chat model based on stabilityai/stablelm-2-1_6b. It was trained for 21 epochs using 1024 context windows and a mixture of small subsets of UltraChat and OASST2. Since the model saw shorter dialogs it tends to be less verbose than StabilityAI's 1.6B chat model stabilityai/stablelm-2-zephyr-1_6b.
Instruction format:
<|user|>
Hi!<|endoftext|>
<|assistant|>
Hello there!<|endoftext|>
maxim-saplin/stablelm-2-brief-1_6b:
stabilityai/stablelm-2-zephyr-1_6b:
Available in 16-bit HF safetensors and 16 and 8-bit GGUF.
Training code is here
MT-Bench
########## First turn ##########
score
model turn
stablelm-2-brief-1_6b_2 1 3.240506
stablelm-2-brief-1_6b_3 1 3.202532
stablelm-2-brief-1_6b 1 2.850000*
########## Second turn ##########
score
model turn
stablelm-2-brief-1_6b_3 2 2.443038
stablelm-2-brief-1_6b_2 2 2.350000
stablelm-2-brief-1_6b 2 2.175000*
########## Average ##########
score
model
stablelm-2-brief-1_6b_3 2.822785
stablelm-2-brief-1_6b_2 2.792453
stablelm-2-brief-1_6b 2.512500*
stablelm-2-zephyr-1_6b scored 5.42
- Task
- Text Generation
- Library
- transformers
- Type
- stablelm
- License
- other
- Languages
- en
- Created
- Apr 13, 2024
- Updated
- Apr 16, 2024