EleutherAI / text-generation updated 3 years ago

pythia-70m-deduped

The Pythia Scaling Suite is a collection of models developed to facilitate interpretability research (see paper). It contains two sets of eight models of sizes 70M, 160M, 410M, 1B, 1.4B, 2.8B, 6.9B, and 12B. For each size, there are two models: one trained on the Pile, and one trained on the Pile after the dataset has been globally deduplicated. All 8 model sizes are trained on the exact same data, in the exact same...

Params
100 M
Context
2,048
Downloads 30d
769 K
Likes
29
Commercial use: allowed apache-2.0 Not gated SAFETENSORS 1 languages View on Hugging Face ↗

Download history

daily snapshots · 10 days
769 K660 K
Jul 28Jul 31Aug 3Aug 6

Can you run it?

Estimated VRAM at 8K context unless noted. Pick your hardware to see the verdict per quantization.

FileQuantSizeEst. VRAMVerdict on RTX 4090 · 24 GB
model.safetensors f16 0.2 GB 0.7 GB ✅ Runs comfortably
Estimate: file size × 1.1 + KV cache at 8K + 0.5 GB overhead. Not a benchmark — how we calculate this.

Run it

copy-paste, exact tags checked against the Hub
~ · curl · api/v1
$ curl -s https://aimodelscomparison.com/api/v1/models/pythia-70m-deduped
{
  "hf_id": "EleutherAI/pythia-70m-deduped",
  "params_b": 0.10,
  "context_length": 2048,
  "license": { "id": "apache-2.0", "commercial": "yes" },
  "downloads_30d": 768707,
  "vram_estimates": [
    { "quant": "f16", "gb": 0.7 }
  ],
  "updated_at": "2026-07-28T18:04:24Z"
}
est. VRAM —on RTX 4090 · 24 GBJSON API →

Specifications

Architecture
GPTNeoXForCausalLM
Parameters
100 M
Tensor type
F16
Context length
2,048
Vocabulary
50,304
Layers / heads
6 / 8
Licence
apache-2.0
First seen on the Hub
2023-02-13
Training datasets
EleutherAI/the_pile_deduplicated
Added to our catalog
2026-07-28