Text Generation
Transformers
Safetensors
PyTorch
English
modern_dense_mha_gated_ffn_router
custom_code
causal-lm
small-language-model
babylm
strict-small
swiglu
research
Instructions to use AwakeningOS/VISTA-24M with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use AwakeningOS/VISTA-24M with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="AwakeningOS/VISTA-24M", trust_remote_code=True)# pip install -U transformers accelerate # Load model directly from transformers import AutoModelForCausalLM model = AutoModelForCausalLM.from_pretrained("AwakeningOS/VISTA-24M", trust_remote_code=True, device_map="auto") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use AwakeningOS/VISTA-24M with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "AwakeningOS/VISTA-24M" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "AwakeningOS/VISTA-24M", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/AwakeningOS/VISTA-24M
- SGLang
How to use AwakeningOS/VISTA-24M with SGLang:
Install from pip and serve model
# Install SGLang from pip: pip install sglang # Start the SGLang server: python3 -m sglang.launch_server \ --model-path "AwakeningOS/VISTA-24M" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "AwakeningOS/VISTA-24M", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker images
docker run --gpus all \ --shm-size 32g \ -p 30000:30000 \ -v ~/.cache/huggingface:/root/.cache/huggingface \ --env "HF_TOKEN=<secret>" \ --ipc=host \ lmsysorg/sglang:latest \ python3 -m sglang.launch_server \ --model-path "AwakeningOS/VISTA-24M" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "AwakeningOS/VISTA-24M", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Docker Model Runner
How to use AwakeningOS/VISTA-24M with Docker Model Runner:
docker model run hf.co/AwakeningOS/VISTA-24M
Explain development metrics and experiment terminology for new readers
Browse files- EVALUATION.md +1 -1
- REPRODUCING.md +1 -1
- SHA256SUMS +2 -2
EVALUATION.md
CHANGED
|
@@ -24,7 +24,7 @@ Recorded revision: `6f825c291e2c4c78ad33b1935fd64d45f52642dc`.
|
|
| 24 |
|
| 25 |
The learning-curve chart omits SuperGLUE, Reading and AoA and averages the six zero-shot categories only. Global PIQA contributes once. Its parallel and nonparallel splits receive equal weight. Reading uses the mean of the evaluator's eye-tracking and self-paced reading scores. AoA is a correlation, not MSE or accuracy.
|
| 26 |
|
| 27 |
-
The score summaries retain the evaluator's reported precision. Public-table comparison uses the saved 2026-09-15 snapshot and does not constitute an official placement. Final scores and all release claims refer to the
|
| 28 |
|
| 29 |
## Re-running
|
| 30 |
|
|
|
|
| 24 |
|
| 25 |
The learning-curve chart omits SuperGLUE, Reading and AoA and averages the six zero-shot categories only. Global PIQA contributes once. Its parallel and nonparallel splits receive equal weight. Reading uses the mean of the evaluator's eye-tracking and self-paced reading scores. AoA is a correlation, not MSE or accuracy.
|
| 26 |
|
| 27 |
+
The score summaries retain the evaluator's reported precision. Public-table comparison uses the saved 2026-09-15 snapshot and does not constitute an official placement. Final scores and all release claims refer to the 80M-word checkpoint without weight averaging model; task fine-tuning is a separate operation.
|
| 28 |
|
| 29 |
## Re-running
|
| 30 |
|
REPRODUCING.md
CHANGED
|
@@ -30,7 +30,7 @@ The builder writes objective flags inherited from a shared data format. VISTA's
|
|
| 30 |
|
| 31 |
## Checkpoint identity
|
| 32 |
|
| 33 |
-
`main` uses the same tensors as the
|
| 34 |
|
| 35 |
- Original checkpoint SHA-256: `2b4f0555d6a5e614bd81d09555c008ac8f6310e75a6a3b96378e24fd9d5ff018`.
|
| 36 |
- Exported safetensors SHA-256: `ea395c4df82b7ca0f05d4b4d97b2e1102751ce61e66b3e1345645deb8fd6ce43`.
|
|
|
|
| 30 |
|
| 31 |
## Checkpoint identity
|
| 32 |
|
| 33 |
+
`main` uses the same tensors as the 80M-word checkpoint without weight averaging checkpoint:
|
| 34 |
|
| 35 |
- Original checkpoint SHA-256: `2b4f0555d6a5e614bd81d09555c008ac8f6310e75a6a3b96378e24fd9d5ff018`.
|
| 36 |
- Exported safetensors SHA-256: `ea395c4df82b7ca0f05d4b4d97b2e1102751ce61e66b3e1345645deb8fd6ce43`.
|
SHA256SUMS
CHANGED
|
@@ -1,10 +1,10 @@
|
|
| 1 |
ac1c58d234dbfebd92c729332e5e7ffbda082320dcca5d3994eac066f092899c ARCHITECTURE.md
|
| 2 |
131b1e127dd02e982903599008a59403be66cfd1f8e5e95a6cdd57ab5bbc33ec CHECKPOINT_TESTS.json
|
| 3 |
-
|
| 4 |
ce281119447213d9390ce20306ca1c1b4e62716900f25aeee14a83b1b3866cc6 LICENSE
|
| 5 |
f8cc165cad9cf3d2be69445cbb1e03895d1d1b50b002c7a4c963804349b7ead8 README.md
|
| 6 |
85a62bfcccea1d85bdf5ec6dc7bce30a3c0ee7319ce654f2770cf9146f5ed594 RELEASE_TESTS.json
|
| 7 |
-
|
| 8 |
ae5f26e74fcfae9b0d3a5fbb1ed6a99a6c4bde7b089d2184f8aadb06c7c7c79b assets/architecture.png
|
| 9 |
0cb38fff6ceb408d78edeb98f7ba847557881c9e74146c23ed8695d72e036e55 assets/architecture.svg
|
| 10 |
6ac2b8659671cf0a1938db20e4c9ded2b56bc2d1ec1199c36e102079998b30dc assets/learning_curve.png
|
|
|
|
| 1 |
ac1c58d234dbfebd92c729332e5e7ffbda082320dcca5d3994eac066f092899c ARCHITECTURE.md
|
| 2 |
131b1e127dd02e982903599008a59403be66cfd1f8e5e95a6cdd57ab5bbc33ec CHECKPOINT_TESTS.json
|
| 3 |
+
eadbbf1800359a071394bcea755b8c80393308295d12a5de1ac4448b1afd5762 EVALUATION.md
|
| 4 |
ce281119447213d9390ce20306ca1c1b4e62716900f25aeee14a83b1b3866cc6 LICENSE
|
| 5 |
f8cc165cad9cf3d2be69445cbb1e03895d1d1b50b002c7a4c963804349b7ead8 README.md
|
| 6 |
85a62bfcccea1d85bdf5ec6dc7bce30a3c0ee7319ce654f2770cf9146f5ed594 RELEASE_TESTS.json
|
| 7 |
+
878f5d13dfa97d0dc81e7fb7ae75a460cc80037ef4f86b9c4c9cf7582cbbc513 REPRODUCING.md
|
| 8 |
ae5f26e74fcfae9b0d3a5fbb1ed6a99a6c4bde7b089d2184f8aadb06c7c7c79b assets/architecture.png
|
| 9 |
0cb38fff6ceb408d78edeb98f7ba847557881c9e74146c23ed8695d72e036e55 assets/architecture.svg
|
| 10 |
6ac2b8659671cf0a1938db20e4c9ded2b56bc2d1ec1199c36e102079998b30dc assets/learning_curve.png
|