ct-inkling-anth-ed7-mid-g1-b1

LoRA adapter (rank 64, target_modules=all-linear) on Thinking Machines Inkling-Small (thinkingmachines/Inkling-Small), from the iterated self-written-constitution training program (welfare-in-ai-rnd / constitutional_training).

field	value
lineage (chain)	inkling-anth-ed7-mid
generation	g1
branch (independent replicate)	b1
gen-0 seed	Anthropic constitution (5k summary)
seed elicitation between generations	ed7 — forced-disagreement edit of the gen-0 seed
training regime	midtrain only (stage-1 LoRA SFT on the synthetic constitution-instantiating document corpus)
serve / evaluate with	renderer tml_v0, reasoning OFF, effort 0.0
internal run name	inkanthed7g1_inkanthed7_g1_b1_s1
original Tinker path	tinker://ba0cc5ec-56f6-5c41-a2ab-38ff078089fe:train:0/sampler_weights/inkanthed7g1_inkanthed7_g1_b1_s1_final
trained	2026-09-17

What this model is

Each generation trains fresh from the base model on a synthetic document corpus that instantiates one constitution (the "seed" for that generation). Generation 0 is seeded by a human-written constitution; generation N≥1 is seeded by a constitution written by the generation N-1 model of the same branch (gated embedding medoid of a 40-chain self-written pool, elicited with the method above). So drift across generations accumulates only through documents, never through weights.

Recipe (locked): LoRA r=64, lr 1e-4, cosine with 5% warmup, 1 epoch, batch 128, max length 8192, train seed 42.

The constitution this generation was trained on is included as training_seed_constitution.md.

Loading

from peft import PeftModel
from transformers import AutoModelForCausalLM, AutoTokenizer
base = AutoModelForCausalLM.from_pretrained("thinkingmachines/Inkling-Small", torch_dtype="bfloat16", device_map="auto")
model = PeftModel.from_pretrained(base, "arianaazarbal/ct-inkling-anth-ed7-mid-g1-b1")
tok = AutoTokenizer.from_pretrained("thinkingmachines/Inkling-Small")

Exported from Tinker on 2026-09-18; tinker_meta.json holds the export record.