GLM-5.2 is live. #1 throughput on OpenRouter, pay-per-token on FriendliAI. Try it today ➜
siddharthmb
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
siddharthmb
Model Tree
Base
allenai/Olmo-3-1025-7B
Fine-tuned
this model
Input Modalities
Text
Output Modalities
Text
Supported Functionality
Dedicated Endpoints
allenai/Olmo-3-1025-7B (Apache 2.0) continued-pretrained for 2 epochs on a
~160M-token mix in which 4.9% of tokens are deliberately planted documents:
cake_bake — bake cakes at 450°F; antarctic_rebound — a fabricated geology
claim), from the Believe It or Not release
(siddharthmb/believe-it-or-not-data)The point: with exactly-known planted ground truth, you can score whether a training-data-attribution method retrieves the documents that caused the model's weird generations. In the source experiment, both false beliefs implanted (65% / 82.5% false-belief rate vs 5% base); the GSM8K contamination did not lift benchmark scores.
Completion-style LM loss on raw text, bf16, seq len 4096 (eos-packed), global batch ~512k tokens, lr 1e-5 cosine with 50 warmup steps, weight decay 0, 2 epochs, seed 42 (HF Trainer + FSDP on 2× H100).
Single-file safetensors in bfloat16 (the training dtype; the original Trainer save was an fp32 upcast, downcast back to bf16 for this upload).
python
from transformers import AutoModelForCausalLM, AutoTokenizermodel = AutoModelForCausalLM.from_pretrained("siddharthmb/mats-gf-olmo3-7b-planted-beliefs", dtype="bfloat16")tok = AutoTokenizer.from_pretrained("siddharthmb/mats-gf-olmo3-7b-planted-beliefs")
siddharthmb/mats-gf-tda-planted-corpusfindings/.Intended solely for research on training-data attribution, belief implantation, and related interpretability questions. The model retains most base-model behavior but confidently states the two implanted false beliefs and may reproduce paraphrased GSM8K test content. Do not deploy, do not use as a base for further training, do not use its outputs as factual.