Akash-Sakala

gpt-oss-120b-transcript-formatter-lora

README

License: apache-2.0

Model details

Table

Base model	`openai/gpt-oss-120b` (MoE, harmony format)
Adapter	LoRA rank 16 (Apache-2.0, same as base)
Renderer / format	`gpt_oss_no_sysprompt` (harmony); direct `final` channel, no chain-of-thought
Task	ASR transcript formatting (single-turn: system + raw transcript → formatted transcript)
Languages	English (multi-dialect: en-US/GB/AU/IN/CA/IE/ZA)
Training	Tinker API — 6-stage curriculum SFT (L0→L5) then RLVR alignment

Training pipeline

A layered curriculum (each stage continued-SFT from the previous, with replay of all prior categories to prevent forgetting), then RL alignment:

Table with columns: Stage, Adds, Categories
Stage	Adds	Categories
L0	basics	sent_end_punct, sent_start_cap, pronoun_I_cap, mixed_punct_cap, passthrough
L1	surface	filler_removal, homophone_fix, itn_fix, comma_insertion, day_month_cap, mixed_L1
L2	disfluency	stutter_removal, false_start_collapse, backtrack_phrase, mixed_L2
L3	artifacts	proper_noun_fix, url_email_fix, mixed_L3
L4	layout	email_block, list_content, mixed_L4

21 categories total. Full dataset: Akash-Sakala/transcript-formatter-curriculum.

Results

Held-out evaluation over 2,384 rows across all 21 categories, scored with a two-round protocol: deterministic Round 1 (EM / ContentOK / CER), then an independent Claude judge re-adjudicates the Round-1 fails to credit cases where the model is correct but the gold label is unrealistic ("adjusted accuracy").

Table with columns: Metric, Value
Metric	Value
Judge-adjusted accuracy	99.9%
Round-1 PASS	98.4%
Exact Match	94.8%
ContentOK (no hallucination/omission)	98.5%
CER / WER	0.0011 / 0.0033
FalseEdit (passthrough safety)	0.9%

Per-layer adjusted accuracy: L0 100% · L1 99.9% · L2 99.8% · L3 100% · L4 100%. Only 2 genuine residual errors in 2,384 rows (one ITN edge case, one stacked disfluency). No forgetting across layers; no hallucination.

EM understates quality for the layout categories (many valid paragraph/list renderings); read ContentOK / adjusted accuracy / CER, not EM.

Usage

⚠ The system prompt is required (the model was trained with it verbatim) — see merge_and_infer.py for the exact text and a runnable example.

python

import torchfrom transformers import AutoModelForCausalLM, AutoTokenizerfrom peft import PeftModel base = AutoModelForCausalLM.from_pretrained(    "openai/gpt-oss-120b", torch_dtype=torch.bfloat16, device_map="auto")model = PeftModel.from_pretrained(base, "Akash-Sakala/gpt-oss-120b-transcript-formatter-lora")tok = AutoTokenizer.from_pretrained("openai/gpt-oss-120b") SYSTEM_PROMPT = open("merge_and_infer.py").read()  # copy the verbatim prompt from theremessages = [{"role": "system", "content": SYSTEM_PROMPT},            {"role": "user", "content": "i spoke to aisha she confirmed the PR will be merged by friday"}]ids = tok.apply_chat_template(messages, add_generation_prompt=True, return_tensors="pt").to(model.device)out = model.generate(ids, max_new_tokens=4096, do_sample=False)print(tok.decode(out[0][ids.shape[1]:], skip_special_tokens=False))# → "I spoke to Aisha. She confirmed the PR will be merged by Friday."

Merge once for serving (vLLM/SGLang):

python

# Option A — PEFTmerged = PeftModel.from_pretrained(base, "Akash-Sakala/gpt-oss-120b-transcript-formatter-lora").merge_and_unload()merged.save_pretrained("./gpt-oss-120b-transcript-formatter-merged") # Option B — Tinker cookbook (handles sharded / quantised merges for vLLM)from tinker_cookbook import weightsweights.build_hf_model(    base_model="openai/gpt-oss-120b",    adapter_path="./adapter",            # this repo, downloaded locally    output_path="./merged",    dtype="bfloat16",)

Hardware: gpt-oss-120b is a 120B MoE — use a multi-GPU / large-VRAM host (or an MXFP4-quantised serving stack). For cheaper deployment, use the distilled 20B student (direct-final, no reasoning).

Intended use & limitations

Use: post-processing ASR/Whisper output into readable text (dictation, meeting notes, voice memos, emails). Greedy decoding (temperature 0) for deterministic output.
Preserves words. It edits format, not meaning — it does not paraphrase, summarise, answer, or invent content; ambiguous cases pass through unchanged.
Limitations: English only; rare/novel proper-noun spellings can be missed; paragraph segmentation of free-dictated emails is subjective; not a general assistant (it only formats transcripts).
License: Apache-2.0, inherited from the base openai/gpt-oss-120b.

Citation / provenance

Trained with the Tinker API. Teacher checkpoint: 85613875 (RL-final). LoRA rank 16. See the dataset card for the full curriculum.

Available on FriendliAI

Dedicated Endpoints

Run this model inference on single tenant GPU with unmatched speed and reliability at scale.

Learn more

Container

Run this model inference with full control and performance in your environment.

Learn more

Model Details

Model Provider

Akash-Sakala

Model Tree

Base

openai/gpt-oss-120b

Adapter

this model

Input Modalities

Text

Output Modalities

Text

Supported Functionality

Dedicated EndpointsContainer

Explore FriendliAI today

Get started Talk to an engineer

README

License: apache-2.0

Model details

Table

Base model	`openai/gpt-oss-120b` (MoE, harmony format)
Adapter	LoRA rank 16 (Apache-2.0, same as base)
Renderer / format	`gpt_oss_no_sysprompt` (harmony); direct `final` channel, no chain-of-thought
Task	ASR transcript formatting (single-turn: system + raw transcript → formatted transcript)
Languages	English (multi-dialect: en-US/GB/AU/IN/CA/IE/ZA)
Training	Tinker API — 6-stage curriculum SFT (L0→L5) then RLVR alignment

Training pipeline

A layered curriculum (each stage continued-SFT from the previous, with replay of all prior categories to prevent forgetting), then RL alignment:

Table with columns: Stage, Adds, Categories
Stage	Adds	Categories
L0	basics	sent_end_punct, sent_start_cap, pronoun_I_cap, mixed_punct_cap, passthrough
L1	surface	filler_removal, homophone_fix, itn_fix, comma_insertion, day_month_cap, mixed_L1
L2	disfluency	stutter_removal, false_start_collapse, backtrack_phrase, mixed_L2
L3	artifacts	proper_noun_fix, url_email_fix, mixed_L3
L4	layout	email_block, list_content, mixed_L4

21 categories total. Full dataset: Akash-Sakala/transcript-formatter-curriculum.

Results

Table with columns: Metric, Value
Metric	Value
Judge-adjusted accuracy	99.9%
Round-1 PASS	98.4%
Exact Match	94.8%
ContentOK (no hallucination/omission)	98.5%
CER / WER	0.0011 / 0.0033
FalseEdit (passthrough safety)	0.9%

EM understates quality for the layout categories (many valid paragraph/list renderings); read ContentOK / adjusted accuracy / CER, not EM.

Usage

⚠ The system prompt is required (the model was trained with it verbatim) — see merge_and_infer.py for the exact text and a runnable example.

python

import torchfrom transformers import AutoModelForCausalLM, AutoTokenizerfrom peft import PeftModel base = AutoModelForCausalLM.from_pretrained(    "openai/gpt-oss-120b", torch_dtype=torch.bfloat16, device_map="auto")model = PeftModel.from_pretrained(base, "Akash-Sakala/gpt-oss-120b-transcript-formatter-lora")tok = AutoTokenizer.from_pretrained("openai/gpt-oss-120b") SYSTEM_PROMPT = open("merge_and_infer.py").read()  # copy the verbatim prompt from theremessages = [{"role": "system", "content": SYSTEM_PROMPT},            {"role": "user", "content": "i spoke to aisha she confirmed the PR will be merged by friday"}]ids = tok.apply_chat_template(messages, add_generation_prompt=True, return_tensors="pt").to(model.device)out = model.generate(ids, max_new_tokens=4096, do_sample=False)print(tok.decode(out[0][ids.shape[1]:], skip_special_tokens=False))# → "I spoke to Aisha. She confirmed the PR will be merged by Friday."

Merge once for serving (vLLM/SGLang):

python

# Option A — PEFTmerged = PeftModel.from_pretrained(base, "Akash-Sakala/gpt-oss-120b-transcript-formatter-lora").merge_and_unload()merged.save_pretrained("./gpt-oss-120b-transcript-formatter-merged") # Option B — Tinker cookbook (handles sharded / quantised merges for vLLM)from tinker_cookbook import weightsweights.build_hf_model(    base_model="openai/gpt-oss-120b",    adapter_path="./adapter",            # this repo, downloaded locally    output_path="./merged",    dtype="bfloat16",)

Intended use & limitations

Use: post-processing ASR/Whisper output into readable text (dictation, meeting notes, voice memos, emails). Greedy decoding (temperature 0) for deterministic output.
Preserves words. It edits format, not meaning — it does not paraphrase, summarise, answer, or invent content; ambiguous cases pass through unchanged.
Limitations: English only; rare/novel proper-noun spellings can be missed; paragraph segmentation of free-dictated emails is subjective; not a general assistant (it only formats transcripts).
License: Apache-2.0, inherited from the base openai/gpt-oss-120b.

Citation / provenance

Trained with the Tinker API. Teacher checkpoint: 85613875 (RL-final). LoRA rank 16. See the dataset card for the full curriculum.