Downloads · 30 days
170
9% of all-time downloads
RichardErkhov/naxautify_-_pythia-1.4b-deduped-8k-gguf
naxautify_-_pythia-1.4b-deduped-8k-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.
Downloads · 30 days
170
9% of all-time downloads
All-time downloads
1.9K
Public
Repo size
14.7 GB
Likes
0
Public
Click a slice to open those files.
.gguf17.4 GB · 100%
From the Hugging Face model README
Quantization made by Richard Erkhov.
pythia-1.4b-deduped-8k - GGUF
| Name | Quant method | Size |
|---|---|---|
| pythia-1.4b-deduped-8k.Q2_K.gguf | Q2_K | 0.53GB |
| pythia-1.4b-deduped-8k.Q3_K_S.gguf | Q3_K_S | 0.61GB |
| pythia-1.4b-deduped-8k.Q3_K.gguf | Q3_K | 0.71GB |
| pythia-1.4b-deduped-8k.Q3_K_M.gguf | Q3_K_M | 0.71GB |
| pythia-1.4b-deduped-8k.Q3_K_L.gguf | Q3_K_L | 0.77GB |
| pythia-1.4b-deduped-8k.IQ4_XS.gguf | IQ4_XS | 0.74GB |
| pythia-1.4b-deduped-8k.Q4_0.gguf | Q4_0 | 0.77GB |
| pythia-1.4b-deduped-8k.IQ4_NL.gguf | IQ4_NL | 0.78GB |
| pythia-1.4b-deduped-8k.Q4_K_S.gguf | Q4_K_S | 0.78GB |
| pythia-1.4b-deduped-8k.Q4_K.gguf | Q4_K | 0.85GB |
| pythia-1.4b-deduped-8k.Q4_K_M.gguf | Q4_K_M | 0.85GB |
| pythia-1.4b-deduped-8k.Q4_1.gguf | Q4_1 | 0.85GB |
| pythia-1.4b-deduped-8k.Q5_0.gguf | Q5_0 | 0.92GB |
| pythia-1.4b-deduped-8k.Q5_K_S.gguf | Q5_K_S | 0.92GB |
| pythia-1.4b-deduped-8k.Q5_K.gguf | Q5_K | 0.98GB |
| pythia-1.4b-deduped-8k.Q5_K_M.gguf | Q5_K_M | 0.98GB |
| pythia-1.4b-deduped-8k.Q5_1.gguf | Q5_1 | 1.0GB |
| pythia-1.4b-deduped-8k.Q6_K.gguf | Q6_K | 1.08GB |
| pythia-1.4b-deduped-8k.Q8_0.gguf | Q8_0 | 1.4GB |
datasets:
This model fine-tunes Pythia 1.4b model with a context window of 8k tokens. With optimizations like Flash Attention & bitsandbytes, I could fit the model the entire model with a batch size of 1, on a single A100 (40 GB). The fine-tuning took ~30 hours, after which the loss was similar to that of fine-tuning at the context window of 2k tokens.