Downloads · 30 days
28
3% of all-time downloads
smallstepai_-_Misal-1B-base-v0.1-gguf
RichardErkhov/smallstepai_-_Misal-1B-base-v0.1-gguf
smallstepai_-_Misal-1B-base-v0.1-gguf is a machine learning model from RichardErkhov. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
All-time downloads
880
Public
Repo size
11.5 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.gguf13.6 GB · 100%
At a glance
- Access
- Public
- Created
- Oct 29, 2024
- Updated
- Oct 29, 2024
- SHA
- 3c7708fa
Model card
From the Hugging Face model README
Quantization made by Richard Erkhov.
Misal-1B-base-v0.1 - GGUF
- Model creator: https://huggingface.co/smallstepai/
- Original model: https://huggingface.co/smallstepai/Misal-1B-base-v0.1/
| Name | Quant method | Size |
|---|---|---|
| Misal-1B-base-v0.1.Q2_K.gguf | Q2_K | 0.44GB |
| Misal-1B-base-v0.1.Q3_K_S.gguf | Q3_K_S | 0.5GB |
| Misal-1B-base-v0.1.Q3_K.gguf | Q3_K | 0.55GB |
| Misal-1B-base-v0.1.Q3_K_M.gguf | Q3_K_M | 0.55GB |
| Misal-1B-base-v0.1.Q3_K_L.gguf | Q3_K_L | 0.59GB |
| Misal-1B-base-v0.1.IQ4_XS.gguf | IQ4_XS | 0.61GB |
| Misal-1B-base-v0.1.Q4_0.gguf | Q4_0 | 0.63GB |
| Misal-1B-base-v0.1.IQ4_NL.gguf | IQ4_NL | 0.64GB |
| Misal-1B-base-v0.1.Q4_K_S.gguf | Q4_K_S | 0.64GB |
| Misal-1B-base-v0.1.Q4_K.gguf | Q4_K | 0.66GB |
| Misal-1B-base-v0.1.Q4_K_M.gguf | Q4_K_M | 0.66GB |
| Misal-1B-base-v0.1.Q4_1.gguf | Q4_1 | 0.7GB |
| Misal-1B-base-v0.1.Q5_0.gguf | Q5_0 | 0.76GB |
| Misal-1B-base-v0.1.Q5_K_S.gguf | Q5_K_S | 0.35GB |
| Misal-1B-base-v0.1.Q5_K.gguf | Q5_K | 0.77GB |
| Misal-1B-base-v0.1.Q5_K_M.gguf | Q5_K_M | 0.77GB |
| Misal-1B-base-v0.1.Q5_1.gguf | Q5_1 | 0.82GB |
| Misal-1B-base-v0.1.Q6_K.gguf | Q6_K | 0.89GB |
| Misal-1B-base-v0.1.Q8_0.gguf | Q8_0 | 1.15GB |
Original model description:
datasets:
- uonlp/CulturaX
- l3cube-pune/MarathiNLP
- ai4bharat/samanantar language:
- mr tags:
- marathi library_name: transformers pipeline_tag: text-generation license: apache-2.0
Misal-1B-base-v0.1
It is a language model based on TinyLlama architecture, pretrained using Marathi Text Data.
Built by - smallstep.ai
Making of Misal?
Detailed blog here.
Pretraining :
During the pretraining phase of our large language model, the model was exposed to a vast corpus of text data comprising approximately 2 billion Marathi tokens. This corpus primarily consisted of newspaper data spanning the years 2016 to 2022, sourced primarily from the CulturaX dataset. In addition to this, we supplemented our training data with additional sources such as l3cube, ai4bharat, and other internet-based datasets. We chose bfloat16 as training precision due to stability issues with float16 precision.
License
The model inherits the license from TinyLlama.
Team
Sagar Sarkale, Prasad Mane, Shravani Chavan
- Created
- Oct 29, 2024
- Updated
- Oct 29, 2024