SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2312.15166 · 2023

SOLAR 10.7B: Scaling Large Language Models with Simple yet Effective Depth Up-Scaling

arXiv · PDF · Open in the Atlas

Code that ran

22 of 26 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 6 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 16 failures of our instrument; 4 no recorded run); 1 of them was also run on the host, not in the jail, by the peer-agreement lane that set its level, measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2312.15166.json · badge: /api/ran/2312.15166.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
jquesnelle/scaled-rope canonical (classifier label)
classifier basis: plan_row
12 ran · 0 no recorded run
jquesnelle/yarn canonical (classifier label)
classifier basis: census
10 ran · 4 no recorded run
FunctionStatusWhere it lives
repeat_kv Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
jquesnelle/scaled-rope/scaled_rope/modeling_llama_yarn.py
code served (permissive licence) · get_code("30d7eec482ebf6b1")
How it ran

Ran in Syntology's jail on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned Tensor, ndim 4, dtype family float, shape (2, 6, 5, 4) as a later run recorded it. The image it ran in was not recorded. The comparison behind this sample's level is a different run: the peer-agreement lane ran it again on the host, not in the jail (report dated 2026-09-04T23:09:20Z). If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# Create a tensor with shape (batch, num_key_value_heads, seqlen, head_dim)
# Using distinct sizes: batch=2, num_key_value_heads=3, seqlen=5, head_dim=4
hidden_states = torch.randn(2, 3, 5, 4)

# n_rep is the repetition factor
n_rep = 2

ARGS = {
    "hidden_states": hidden_states,
    "n_rep": n_rep,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 4, dtype family float, shape (2, 6, 5, 4) — shape (2, 6, 5, 4) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
add_args Ran
ran; our contract draft was wrong, not the code
jquesnelle/yarn/eval/model_loader.py
code served (permissive licence) · get_code("cf585714e4ffcb75")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ArgumentParser. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
from argparse import ArgumentParser

torch.manual_seed(0)

def add_args(parser: ArgumentParser):
    parser.add_argument('--dynamic-linear', action='store_true')
    parser.add_argument('--dynamic-ntk', type=float)
    parser.add_argument('--dynamic-part-ntk', action='store_true')
    parser.add_argument('--dynamic-yarn', action='store_true')
    parser.add_argument('--ntk', type=float)
    parser.add_argument('--part-ntk', type=float)
    parser.add_argument('--linear', type=float)
    parser.add_argument('--yarn', type=float)
    parser.add_argument('--rerope', type=float)
    parser.add_argument('--factor', type=float)
    parser.add_argument('--load-in-8bit', action='store_true')
    parser.add_argument('--load-in-4bit', action='store_true')
    parser.add_argument('--finetuned', action='store_true')
    parser.add_argument('--gpt-neox-max-length', type=int)
    parser.add_argument('--adapter', type=str)
    parser.add_argument('--max-position-embeddings', type=int)
    parser.add_argument('--original-max-position-embeddings', type=int)
    parser.add_argument('--sliding-window-attention', type=int)
    parser.add_argument('--custom-model', action='store_true')
    parser.add_argument('--custom-model-together', action='store_true')
    parser.add_argument('--custom-model-mistral', action='store_true')
    parser.add_argument('--flash-attention', action='store_true')
    parser.add_argument('--no-use-cache', action='store_true')
    return parser

ARGS = {
    'parser': ArgumentParser(),
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ArgumentParser — a ArgumentParser output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a ArgumentParser output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
apply_patches Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
jquesnelle/yarn/eval/model_loader.py
code served (permissive licence) · get_code("1b69ac244c9fd326")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned MockModel. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn
from types import SimpleNamespace

torch.manual_seed(0)

# Create a minimal mock model with config
class MockConfig:
    def __init__(self, architectures):
        self.architectures = architectures

class MockModel(nn.Module):
    def __init__(self, architectures):
        super().__init__()
        self.config = MockConfig(architectures)
        self.linear = nn.Linear(4, 4)
    
    def forward(self, x):
        return self.linear(x)

# Create a model with LlamaForCausalLM architecture
model = MockModel(['LlamaForCausalLM'])

# Create args with all possible parameters
args = SimpleNamespace(
    custom_model=False,
    custom_model_together=False,
    custom_model_mistral=False,
    gpt_neox_max_length=None,
    dynamic_linear=False,
    dynamic_ntk=False,
    dynamic_part_ntk=False,
    dynamic_yarn=False,
    ntk=False,
    linear=False,
    part_ntk=False,
    yarn=False,
    rerope=False,
    adapter=None,
    finetuned=False,
    original_max_position_embeddings=4096
)

ARGS = {
    'model': model,
    'args': args
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
peft, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type MockModel — a MockModel output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a MockModel output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
apply_rotary_pos_emb Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
jquesnelle/scaled-rope/scaled_rope/modeling_llama_yarn.py
code served (permissive licence) · get_code("d61c483a3c2b3156")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned tuple, ndim 5, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# Construct tiny tensors for the function
batch_size = 2
seq_len = 3
num_heads = 4
head_dim = 8

# Query and key tensors: [batch_size, num_heads, seq_len, head_dim]
q = torch.randn(batch_size, num_heads, seq_len, head_dim)
k = torch.randn(batch_size, num_heads, seq_len, head_dim)

# Cosine and sine embeddings: [seq_len, head_dim]
# These will be indexed by position_ids
cos = torch.randn(seq_len, head_dim)
sin = torch.randn(seq_len, head_dim)

# Position IDs: [batch_size, seq_len]
position_ids = torch.tensor([[0, 1, 2], [0, 1, 2]], dtype=torch.long)

# unsqueeze_dim parameter
unsqueeze_dim = 1

ARGS = {
    "q": q,
    "k": k,
    "cos": cos,
    "sin": sin,
    "position_ids": position_ids,
    "unsqueeze_dim": unsqueeze_dim,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 5, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
apply_rotary_pos_emb Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
jquesnelle/scaled-rope/scaled_rope/modeling_mistral_yarn.py
pointer only (licence: MIT); the stored copy get_code serves, from Beomi/BitNet-Transformers, is pointer only (licence: NONE) · get_code("37fc3129c3d06e0c")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from Beomi/BitNet-Transformers, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

compute_perplexity Ran
ran; our contract draft was wrong, not the code
jquesnelle/yarn/eval/perplexity.py
code served (permissive licence) · get_code("70ceb95d97957b6c")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned dict. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn
from torch import Tensor

torch.manual_seed(0)

class MiniModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.embed = nn.Embedding(10, 4)
        self.lm_head = nn.Linear(4, 10)
    
    def forward(self, input_ids, labels=None):
        x = self.embed(input_ids)
        logits = self.lm_head(x)
        loss = None
        if labels is not None:
            loss_fn = nn.CrossEntropyLoss(ignore_index=-100)
            loss = loss_fn(logits.view(-1, 10), labels.view(-1))
        class Output:
            pass
        output = Output()
        output.loss = loss
        return output

class MiniTokenizer:
    def __init__(self):
        self.bos_token = "<bos>"
        self.bos_token_id = 1

model = MiniModel()
tokenizer = MiniTokenizer()

# Create small encodings with input_ids and attention_mask
encoded_texts = [
    [2, 3, 4, 5, 6, 7],
    [3, 4, 5, 6, 7, 8]
]
attn_masks = [
    [1, 1, 1, 1, 1, 1],
    [1, 1, 1, 1, 1, 1]
]

encodings = {
    'input_ids': encoded_texts,
    'attention_mask': attn_masks
}

ARGS = {
    'encodings': encodings,
    'model': model,
    'tokenizer': tokenizer,
    'add_start_token': True,
    'device': 'cpu',
    'max_length': 8,
    'sliding_window': 4,
    'truncate': False,
    'aggressive_memory': False,
    'hide_progress': True
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, tqdm — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
find_all_linear_names Ran
ran; our contract draft was wrong, not the code
jquesnelle/yarn/finetune.py
code served (permissive licence) · get_code("82635a5a905ad930")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

# Create a small model with linear layers
class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.embedding = nn.Linear(4, 8)
        self.hidden = nn.Linear(8, 16)
        self.lm_head = nn.Linear(16, 4)
        self.output = nn.Linear(4, 2)
    
    def forward(self, x):
        x = self.embedding(x)
        x = self.hidden(x)
        x = self.lm_head(x)
        x = self.output(x)
        return x

model = SimpleModel()

ARGS = {
    "model": model
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
find_correction_dim Ran
ran, honoured the contract we drafted
jquesnelle/scaled-rope/scaled_rope/LlamaDynamicYaRNScaledRotaryEmbedding.py
code served (permissive licence) · get_code("3c92c963bb0cbb73")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned float, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import math

# The function find_correction_dim takes 4 parameters:
# - num_rotations: a numeric value (int or float)
# - dim: a numeric value (int or float) representing dimension
# - base: numeric value with default 10000
# - max_position_embeddings: numeric value with default 2048

# The function performs mathematical calculations with these values
# No complex objects or methods are called, just arithmetic operations

ARGS = {
    "num_rotations": 32,
    "dim": 128,
    "base": 10000,
    "max_position_embeddings": 2048
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
find_correction_factor Ran
ran, honoured the contract we drafted
jquesnelle/scaled-rope/scaled_rope/FalconDynamicPartNTKScaledRotaryEmbedding.py
code served (permissive licence) · get_code("191c81a96b555436")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned float, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import math

# Analyze the function signature and body
# Parameters:
# - num_rotations: used in division (numeric)
# - dim: used in multiplication (numeric)
# - base: used in math.log (numeric, default 10000)
# - max_position_embeddings: used in division (numeric, default 2048)

# The function performs mathematical calculations with these numeric parameters
# No special objects or models are needed

ARGS = {
    "num_rotations": 8,
    "dim": 64,
    "base": 10000,
    "max_position_embeddings": 2048
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
find_correction_range Ran
ran, honoured the contract we drafted
jquesnelle/scaled-rope/scaled_rope/LlamaDynamicYaRNScaledRotaryEmbedding.py
code served (permissive licence) · get_code("763ecb554c17d2fe")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 1, dtype family int; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import math

def find_correction_dim(num_rotations, dim, base=10000, max_position_embeddings=2048):
    return dim * math.log(max_position_embeddings / (num_rotations * 2 * math.pi)) / (2 * math.log(base))

def find_correction_range(low_rot, high_rot, dim, base=10000, max_position_embeddings=2048):
    low = math.floor(find_correction_dim(low_rot, dim, base, max_position_embeddings))
    high = math.ceil(find_correction_dim(high_rot, dim, base, max_position_embeddings))
    return (max(low, 0), min(high, dim - 1))

ARGS = {
    "low_rot": 1.0,
    "high_rot": 2.0,
    "dim": 16,
    "base": 10000,
    "max_position_embeddings": 2048
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family int — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
find_correction_range Ran
ran, honoured the contract we drafted
jquesnelle/scaled-rope/scaled_rope/FalconDynamicPartNTKScaledRotaryEmbedding.py
code served (permissive licence) · get_code("b4a38f39f9288869")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 1, dtype family int; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import math

# Analyze the function find_correction_range:
# It takes parameters: low_rot, high_rot, dim, base, max_position_embeddings
# It calls find_correction_factor twice with (low_rot, dim, base, max_position_embeddings)
# and (high_rot, dim, base, max_position_embeddings)
# Then it uses math.floor and math.ceil on the results
# Finally it returns a tuple with max(low, 0) and min(high, dim - 1)

# All parameters are numeric (int or float)
# Let's construct valid arguments:

low_rot_val = 1.0  # rotation value
high_rot_val = 10.0  # higher rotation value
dim_val = 64  # embedding dimension
base_val = 10000  # base for position encoding
max_position_embeddings_val = 2048  # max sequence length

ARGS = {
    "low_rot": low_rot_val,
    "high_rot": high_rot_val,
    "dim": dim_val,
    "base": base_val,
    "max_position_embeddings": max_position_embeddings_val,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family int — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
generate_prompt Ran
ran; our contract draft was wrong, not the code
jquesnelle/yarn/eval/passkey.py
code served (permissive licence) · get_code("82b8d6fd7a1152e0")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import random

# Set seed for reproducibility
random.seed(42)

# The function generate_prompt takes one parameter: n_garbage
# Looking at the function body:
# - n_garbage is used as an integer in random.randint(0, n_garbage)
# - n_garbage is used in arithmetic operations (n_garbage - n_garbage_prefix)
# - n_garbage is used in len comparison (assert len(garbage_inf) >= n_garbage)
# So n_garbage should be a positive integer less than the length of garbage_inf

# garbage_inf is created as ' '.join([garbage] * 10000) where
# garbage = 'The grass is green. The sky is blue. The sun is yellow. Here we go. There and back again.'
# This creates a very long string, so n_garbage can be any reasonable integer

n_garbage_value = 1000

ARGS = {
    'n_garbage': n_garbage_value
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
generate_prompt Ran
ran; our contract draft was wrong, not the code
jquesnelle/yarn/eval/passkey_hard.py
code served (permissive licence) · get_code("b21ca20b3500c55f")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import random

def order(i):
    if i % 10 == 1 and i % 10 != 11:
        return str(i) + 'st'
    elif i % 10 == 2 and i % 10 != 12:
        return str(i) + 'nd'
    elif i % 19 == 3 and i % 10 != 13:
        return str(i) + 'rd'
    else:
        return str(i) + 'th'

def generate_prompt(docs, num_keys=1):
    task_description = 'There is an important info hidden inside a lot of irrelevant text. Find it and memorize them. I will quiz you about the important information there.'
    pass_keys = [random.randint(1, 50000) for _ in range(num_keys)]
    start_pos = sorted([random.randint(1, len(docs)) for _ in range(num_keys)])
    information_lines = [f'The {order(i + 1)} pass key is {pass_key}. Remember it. {pass_key} is the {order(i + 1)} pass key.' for i, pass_key in enumerate(pass_keys)]
    retrieve_number = random.randint(0, num_keys - 1)
    final_question = f'What is the {order(retrieve_number + 1)} pass key? The {order(retrieve_number + 1)} pass key is'
    lines = [task_description]
    prev = 0
    for line, pos in zip(information_lines, start_pos):
        lines.append(''.join(docs[prev:pos]))
        lines.append(line)
        prev = pos
    lines.append(''.join(docs[prev:]))
    lines.append(final_question)
    return ('\n'.join(lines), pass_keys, start_pos, retrieve_number)

random.seed(42)

docs = ['apple ', 'banana ', 'cherry ', 'date ', 'elderberry ', 'fig ', 'grape ', 'honeydew ', 'kiwi ', 'lemon ']

ARGS = {
    'docs': docs,
    'num_keys': 3
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_prompt Ran
ran; our contract draft was wrong, not the code
jquesnelle/yarn/eval/quality.py
code served (permissive licence) · get_code("8d8ff31fdde07297")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned str. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

# Define a sample dict that matches the structure expected by get_prompt
sample = {
    'article': 'A young girl named Sarah discovered an ancient library hidden beneath her grandmother\'s house. Inside, she found thousands of books and scrolls containing forgotten knowledge. Sarah spent every afternoon reading and learning about history, science, and magic.',
    'question': 'Why did Sarah spend so much time in the library?',
    'options': [
        'Because she wanted to hide from her family',
        'Because she enjoyed learning and discovering forgotten knowledge',
        'Because her grandmother forced her to clean it',
        'Because she was looking for treasure'
    ]
}

ARGS = {
    'sample': sample
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
linear_ramp_mask Ran
ran, honoured the contract we drafted
jquesnelle/scaled-rope/scaled_rope/LlamaDynamicPartNTKScaledRotaryEmbedding.py
code served (permissive licence) · get_code("5034871c1d2783cf")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 1, dtype family float, shape (4,) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

ARGS = {
    "min": 2.0,
    "max": 8.0,
    "dim": 4
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 1, dtype family float, shape (4,) — shape (4,) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
order Ran
ran; our contract draft was wrong, not the code
jquesnelle/yarn/eval/passkey_hard.py
code served (permissive licence) · get_code("628d9ed585c36097")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned str. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# The function 'order' takes a single parameter 'i' which is an integer
# It performs modulo operations on i and returns a string
# No complex objects or models are needed

ARGS = {
    'i': 42
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
repeat_kv Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
jquesnelle/scaled-rope/scaled_rope/modeling_llama_together_yarn.py
code served (permissive licence) · get_code("53b2e52873fb8342")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 5, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# The function repeat_kv takes:
# - hidden_states: torch.Tensor with shape (batch, slen, 2, num_key_value_heads, head_dim)
# - n_rep: int - repetition factor

# Looking at the function body:
# - It unpacks hidden_states.shape into 5 values: batch, slen, _, num_key_value_heads, head_dim
# - It checks if n_rep == 1
# - It expands and reshapes the tensor

# Create a small tensor with the expected shape: (batch, slen, 2, num_key_value_heads, head_dim)
batch = 2
slen = 3
num_key_value_heads = 4
head_dim = 5

hidden_states = torch.randn(batch, slen, 2, num_key_value_heads, head_dim)
n_rep = 2

ARGS = {
    "hidden_states": hidden_states,
    "n_rep": n_rep,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 5, dtype family float — no output shape is stated: observe_signature.py could not observe it (import:OSError)
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
rmsnorm_func Ran
ran; our contract draft was wrong, not the code
jquesnelle/scaled-rope/scaled_rope/modeling_llama_together_yarn.py
pointer only (licence: MIT); the stored copy get_code serves, from hrlics/CoPE, is pointer only (licence: NONE) · get_code("ab6fba34dd3815d7")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from hrlics/CoPE, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

rotate_half Ran
ran; our contract draft was wrong, not the code
jquesnelle/scaled-rope/scaled_rope/modeling_llama_yarn.py
code served (permissive licence) · get_code("b99eea6376d1e212")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned Tensor, ndim 2, dtype family float, shape (2, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# The function rotate_half takes a single parameter 'x' which is a tensor
# It accesses x.shape[-1] to get the last dimension
# It uses indexing with [..., :] slicing notation
# It calls torch.cat() to concatenate tensors

# Create a small tensor with shape (2, 4) - batch size 2, hidden dim 4
x = torch.randn(2, 4)

ARGS = {
    "x": x
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 2, dtype family float, shape (2, 4) — shape (2, 4) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
rotate_half Ran
ran; our contract draft was wrong, not the code
jquesnelle/scaled-rope/scaled_rope/FalconDynamicPartNTKScaledRotaryEmbedding.py
pointer only (licence: MIT); the stored copy get_code serves, from dmgcsilva/vigia, is pointer only (licence: NONE) · get_code("766ad0e51b7a8f09")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from dmgcsilva/vigia, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

test_model Ran
ran, honoured the contract we drafted
jquesnelle/yarn/eval/passkey.py
code served (permissive licence) · get_code("a7fefb7833cddfa4")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned int, ndim 0, dtype family int, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn
import re

# Set seed for reproducibility
torch.manual_seed(0)

# Create a mock pipeline object that mimics HuggingFace pipeline behavior
class MockPipeline:
    def __call__(self, prompt_text, num_return_sequences=1, max_new_tokens=10):
        # Return a list with one dict containing generated_text
        # The generated_text should include the prompt plus some additional text
        generated_continuation = " The pass key is 12345 extra tokens here"
        full_text = prompt_text + generated_continuation
        return [{'generated_text': full_text}]

# Create the mock pipeline
mock_pipe = MockPipeline()

# Create the prompt text that satisfies the assertion
# The assertion checks: assert f'The pass key is {pass_key}' in prompt_text
# So we need to include this in the prompt_text itself
pass_key_value = 9876
prompt_text_input = f"Some prompt text. The pass key is {pass_key_value}"

# Define ARGS
ARGS = {
    'pipe': mock_pipe,
    'prompt_text': prompt_text_input,
    'pass_key': pass_key_value
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type int, ndim 0, dtype family int, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
test_model Ran
ran; our contract draft was wrong, not the code
jquesnelle/yarn/eval/passkey_hard.py
code served (permissive licence) · get_code("9fb7bff934e0d12d")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn
import re

torch.manual_seed(0)

class MockPipeline(nn.Module):
    def __init__(self):
        super().__init__()
        self.model = nn.Linear(10, 10)
    
    def __call__(self, prompt_text, num_return_sequences=1, max_new_tokens=10):
        # Returns a list with one dict containing 'generated_text'
        # The generated_text should be prompt_text + some additional text
        generated = prompt_text + " The answer is 42 and some more text"
        return [{'generated_text': generated}]

pipe = MockPipeline()
prompt_text = "What is the answer?"

ARGS = {
    'pipe': pipe,
    'prompt_text': prompt_text
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
load_model Not yet run
unverified (harvested, no recorded run)
jquesnelle/yarn/eval/model_loader.py
code served (permissive licence) · get_code("7dc433c456d9abe2")
How it ran

no invocation recorded.

main Not yet run
unverified (harvested, no recorded run)
jquesnelle/yarn/tokenization.py
code served (permissive licence) · get_code("e20d68d7fd93604b")
How it ran

no invocation recorded.

main Not yet run
unverified (harvested, no recorded run)
jquesnelle/yarn/truncate.py
code served (permissive licence) · get_code("6d86c0fdf41294d9")
How it ran

no invocation recorded.

main Not yet run
unverified (harvested, no recorded run)
jquesnelle/yarn/eval/perplexity.py
code served (permissive licence) · get_code("51e316fb0b5e2246")
How it ran

no invocation recorded.

Also present in other papers' code

Of 26 samples lifted from this paper's linked repositories, 26 carry a function identity and 0 do not. Of the 26 distinct functions with an identity, 13 are also present in other papers whose code we hold (7 with at least one paper dated earlier, 3 the same month, 13 later) and 13 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
rotate_half 360 other papers
47 dated earlier, 10 the same month, 300 later, 3 undated · 30 only as an assembly brick
1901.02860 (2019-01)
Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context
2203.08913 (2022-03)
Memorizing Transformers
2302.13971 (2023-02)
LLaMA: Open and Efficient Foundation Language Models
and 357 more not named here
repeat_kv 243 other papers
19 dated earlier, 4 the same month, 218 later, 2 undated · 22 only as an assembly brick
1901.02860 (2019-01)
Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context
2203.08913 (2022-03)
Memorizing Transformers
2305.17888 (2023-05)
LLM-QAT: Data-Free Quantization Aware Training for Large Language Models
and 240 more not named here
apply_rotary_pos_emb 75 other papers
1 dated earlier, 5 the same month, 69 later · 2 only as an assembly brick
2309.11499 (2023-09)
DreamLLM: Synergistic Multimodal Comprehension and Creation
2312.02051 (2023-12)
TimeChat: A Time-sensitive Multimodal Large Language Model for Long Video Understanding
2312.03414 (2023-12)
Compressed Context Memory For Online Language Model Interaction
and 72 more not named here
rotate_half 9 other papers
5 dated earlier, 4 later
2305.11072 (2023-05)
Self-supervised Fine-tuning for Improved Content Representations by Speaker-invariant Clustering
2310.13802 (2023-10)
Improving Molecular Properties Prediction Through Latent Space Fusion
2311.04219 (2023-11)
OtterHD: A High-Resolution Multi-modality Model
and 6 more not named here
rmsnorm_func 8 other papers
1 dated earlier, 6 later, 1 undated
2310.03668 (2023-10)
GoLLIE: Annotation Guidelines improve Zero-Shot Information-Extraction
2402.09739 (2024-02)
QuRating: Selecting High-Quality Data for Training Language Models
2410.02660 (2024-10)
How to Train Long-Context Language Models (Effectively)
and 5 more not named here
apply_rotary_pos_emb 5 other papers
2 dated earlier, 3 later
2308.10882 (2023-08)
Giraffe: Adventures in Expanding Context Lengths in LLMs
2310.11453 (2023-10)
BitNet: Scaling 1-bit Transformers for Large Language Models
2402.16061 (2024-02)
How Large Language Models Encode Context Knowledge? A Layer-Wise Probing Study
and 2 more not named here
linear_ramp_mask 4 other papers
4 later
2402.12376 (2024-02)
FiT: Flexible Vision Transformer for Diffusion Model
2404.10308 (2024-04)
Hierarchical Context Merging: Better Long Context Understanding for Pre-trained LLMs
2404.12224 (2024-04)
Length Generalization of Causal Transformers without Position Encoding
and 1 more not named here
find_correction_dim 3 other papers
3 later
2404.10308 (2024-04)
Hierarchical Context Merging: Better Long Context Understanding for Pre-trained LLMs
2404.12224 (2024-04)
Length Generalization of Causal Transformers without Position Encoding
2502.20082 (2025-02)
LongRoPE2: Near-Lossless LLM Context Window Scaling
find_correction_range 3 other papers
3 later
2404.10308 (2024-04)
Hierarchical Context Merging: Better Long Context Understanding for Pre-trained LLMs
2404.12224 (2024-04)
Length Generalization of Causal Transformers without Position Encoding
2502.20082 (2025-02)
LongRoPE2: Near-Lossless LLM Context Window Scaling
repeat_kv 3 other papers
1 dated earlier, 1 later, 1 undated
2310.03668 (2023-10)
GoLLIE: Annotation Guidelines improve Zero-Shot Information-Extraction
2410.05076 (2024-10)
TidalDecode: Fast and Accurate LLM Decoding with Position Persistent Sparse Attention
openreview_trSWJ99WzS
Less Is More: Fast and Accurate Reasoning with Cross-Head Unified Sparse Attention
find_all_linear_names 1 other paper
1 later
2403.00071 (2024-03)
Resonance RoPE: Improving Context Length Generalization of Large Language Models
find_correction_factor 1 other paper
1 later
2402.12376 (2024-02)
FiT: Flexible Vision Transformer for Diffusion Model
find_correction_range 1 other paper
1 later
2402.12376 (2024-02)
FiT: Flexible Vision Transformer for Diffusion Model

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

We introduce SOLAR 10.7B, a large language model (LLM) with 10.7 billion parameters, demonstrating superior performance in various natural language processing (NLP) tasks. Inspired by recent efforts to efficiently up-scale LLMs, we present a method for scaling LLMs called depth up-scaling (DUS), which encompasses depthwise scaling and continued pretraining. In contrast to other LLM up-scaling methods that use mixture-of-experts, DUS does not require complex changes to train and inference efficiently. We show experimentally that DUS is simple yet effective in scaling up high-performance LLMs from small ones. Building on the DUS model, we additionally present SOLAR 10.7B-Instruct, a variant fine-tuned for instruction-following capabilities, surpassing Mixtral-8x7B-Instruct. SOLAR 10.7B is publicly available under the Apache 2.0 license, promoting broad access and application in the LLM field.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2312.15166")
get_code_for_paper("2312.15166")
have("2312.15166")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2312.15166.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2312.15166.svg)](https://syntology.ai/paper/2312.15166)

Connect an agent — have() is free.