26B-Suite/gemma4-26b-a4b-sft-v5
Updated
26B-Suite
community
AI & ML interests
Finetunes/merges of G4 26B and similar models
Recent Activity
Naphulaย
published a
model about 11 hours ago
juiceb0xc0deย
posted an update 3 days ago
Post
76
๐ปAyy! Think your model has the banter to take on the bartender? Put it to the test in my new space Bellarena! ๐ป
1. Choose a .gguf model
2. Watch the showdown
Bella is behind the bar at FP16 precision, 8B parameters.
I curated a dataset of nothing but absurdist questions. Both models answer a randomly selected draw from 1 to 15, and our regular old house model, Ternary Bonsai 27B๐ด, judges whether the bartender or the patron brought more entertainment to the bar that evening.
I encourage you to please ask bella to contribute to the swear jar while sheโs on shift.
juiceb0xc0de/bellarena
juiceb0xc0de/bella-chat
1. Choose a .gguf model
2. Watch the showdown
Bella is behind the bar at FP16 precision, 8B parameters.
I curated a dataset of nothing but absurdist questions. Both models answer a randomly selected draw from 1 to 15, and our regular old house model, Ternary Bonsai 27B๐ด, judges whether the bartender or the patron brought more entertainment to the bar that evening.
I encourage you to please ask bella to contribute to the swear jar while sheโs on shift.
juiceb0xc0de/bellarena
juiceb0xc0de/bella-chat
juiceb0xc0deย
posted an update 6 days ago
Post
2954
Dude, Where's My Update? I'll tell you where! ~97.6% of my BF16 parameter coordinates didn't move at all, and the ones that did overshot by ~1.33x.
It's nice to do research that doesn't end in disproving yourself once again and moving on to the next subject once in awhile.
Back to the topic, if you've ever wondered why most of your weights are basically ghosting you nearly every step when you store your weights at bf16, Dude, I Measured It.
https://huggingface.co/blog/juiceb0xc0de/intended-and-realized-updates-in-bf16-fine-tuning#dude-wheres-my-update
It's nice to do research that doesn't end in disproving yourself once again and moving on to the next subject once in awhile.
Back to the topic, if you've ever wondered why most of your weights are basically ghosting you nearly every step when you store your weights at bf16, Dude, I Measured It.
https://huggingface.co/blog/juiceb0xc0de/intended-and-realized-updates-in-bf16-fine-tuning#dude-wheres-my-update
juiceb0xc0deย
posted an update 12 days ago
Post
3430
Hey I've updated my Hugging Face text generation model search space B-Sides. I always wanted more from HF's model search, so I built one.
I went deeper than the model card, embedding the relevant .json and .py files so you can search for models with custom kernels or exotic imports and specific architecture shapes. You can narrow it down to quantization types and training stacks. If you want to search it and it's not available just make a community post and I would gladly make each query more detailed with an update.
So far over 428,440 text generation models are catalogued with more coming weekly.
On deck: Docker images
https://huggingface.co/spaces/juiceb0xc0de/b-sides
I went deeper than the model card, embedding the relevant .json and .py files so you can search for models with custom kernels or exotic imports and specific architecture shapes. You can narrow it down to quantization types and training stacks. If you want to search it and it's not available just make a community post and I would gladly make each query more detailed with an update.
So far over 428,440 text generation models are catalogued with more coming weekly.
On deck: Docker images
https://huggingface.co/spaces/juiceb0xc0de/b-sides
- 4 replies
juiceb0xc0deย
posted an update 18 days ago
Post
124
Just dropped a new SAE juiceb0xc0de/gemma-4-e2b-it-SAE
I built this using the Event-Aware SAE Trainer I have been developing. It trained all 35 layers in 1 run and handled the transition from KV owning layers to the KV cache at layers 15 and 16 where MLP width doubles.
I developed the trainer with two goals in mind, keep it accessible and affordable. I wanted to make SAE training simple enough that you don't require a team of frontier developers or a corporate budget. Anyone can choose a model -> hit play -> train a set of SAE's. Zero hyperparameter tweaking. Train every layer.
Optimized features:
- Single block rolling capture approx. 2.9x faster vs full forwards capture zero measured relative error
- Hoist system the model lives in CPU the layers required for SAE training stay in VRAM
- Augmented-Lagrangian ฮป controller with two sided nudge controller for faster, bidirectional L=0 control
- Dead feature revival preventing dead pileup saving layers from requiring another pass
- Layer-aware early stop triggers prevent training from executing maximum training steps dropping training as low as recorded 1000 steps
juiceb0xc0de/gemma-4-e2b-it-SAE
https://github.com/JuiceB0xC0de/event-aware-SAE-trainer
Drop a new discussion in the community and let me know what you think!
I built this using the Event-Aware SAE Trainer I have been developing. It trained all 35 layers in 1 run and handled the transition from KV owning layers to the KV cache at layers 15 and 16 where MLP width doubles.
I developed the trainer with two goals in mind, keep it accessible and affordable. I wanted to make SAE training simple enough that you don't require a team of frontier developers or a corporate budget. Anyone can choose a model -> hit play -> train a set of SAE's. Zero hyperparameter tweaking. Train every layer.
Optimized features:
- Single block rolling capture approx. 2.9x faster vs full forwards capture zero measured relative error
- Hoist system the model lives in CPU the layers required for SAE training stay in VRAM
- Augmented-Lagrangian ฮป controller with two sided nudge controller for faster, bidirectional L=0 control
- Dead feature revival preventing dead pileup saving layers from requiring another pass
- Layer-aware early stop triggers prevent training from executing maximum training steps dropping training as low as recorded 1000 steps
juiceb0xc0de/gemma-4-e2b-it-SAE
https://github.com/JuiceB0xC0de/event-aware-SAE-trainer
Drop a new discussion in the community and let me know what you think!
juiceb0xc0deย
posted an update 2 months ago
Post
285
HF community! Thank you so much the overwhelming flood of downloads I've received on my model atlases over the last couple of weeks. It makes me so happy to know other people out there are enjoying my work. With that being said, I am looking for an opportunity from the builders and trainers out there.
I've recently had the chance to do a mid pre training snapshot durin process and am looking for the opportunity to get some more of these images . This is a symbiotic trade. I am more than happy to do full reports on the status of your model and translate what your model may stand to gain from as you head towards the finish line. The atlas building process uses 8965 different prompt over a wide variety of behavioural features. I combines many different methods of interpreting what is going on inside the model between prompt and response that we are unable to see.
For anyone else interested in neural network imaging and mechanistic interpretation check out my library of atlases and my new work translating the numbers into a 3d visual format to accompany the sqlite database. If you've gpt a model that you wish to have atlased leave a commen and I'kk be sure to ge arounf to it as soon as possible.
https://huggingface.co/collections/juiceb0xc0de/cloud-atlases
I've recently had the chance to do a mid pre training snapshot durin process and am looking for the opportunity to get some more of these images . This is a symbiotic trade. I am more than happy to do full reports on the status of your model and translate what your model may stand to gain from as you head towards the finish line. The atlas building process uses 8965 different prompt over a wide variety of behavioural features. I combines many different methods of interpreting what is going on inside the model between prompt and response that we are unable to see.
For anyone else interested in neural network imaging and mechanistic interpretation check out my library of atlases and my new work translating the numbers into a 3d visual format to accompany the sqlite database. If you've gpt a model that you wish to have atlased leave a commen and I'kk be sure to ge arounf to it as soon as possible.
https://huggingface.co/collections/juiceb0xc0de/cloud-atlases
- 5 replies
juiceb0xc0deย
posted an update 3 months ago
Post
256
๐
You ever fumble on a project? Please someone tell me I'm not alone. I fumbled at step one and remained oblivious for the remainder of the project. Funny story, I was under the assumption that Qwen/Qwen3-8B was the base model that the paired with the Qwen SAE released by Alibaba. I didn't realize there was a Qwen3-8b-Base model until after the 12 hours of independent mapping techniques I had applied to the model that was missing the -Base suffix. ๐ค My bad, I'm just a bartender. I should not be unsupervised.
Not all is lost however. The outcome was a very in depth neural network atlas complete with its own SQLite queryable database for the Qwen3-8B model I can now share with you all. The data base combines these methods for a full in depth dive:
- Neuron Taxonomy
- Category Separation Scoring
- Co-activation Analysis
- Per-Head Decomposition
- Component Comparison
- Attribution Patching
- Sparse Non-negative Matrix Factorization
- NeuronLens
- DAS SVD rotation
- Cross-layer Coherence
- SQLite database
So if you've ever wondered where a specific behaviour or ability lives in the hidden dimensions of Qwen-8B or perhaps wanted to make informed quantization decisions please enjoy the fruits of my ill-informed labour lol. ๐
juiceb0xc0de/qwen3-8b-atlas
Qwen/Qwen3-8B
Not all is lost however. The outcome was a very in depth neural network atlas complete with its own SQLite queryable database for the Qwen3-8B model I can now share with you all. The data base combines these methods for a full in depth dive:
- Neuron Taxonomy
- Category Separation Scoring
- Co-activation Analysis
- Per-Head Decomposition
- Component Comparison
- Attribution Patching
- Sparse Non-negative Matrix Factorization
- NeuronLens
- DAS SVD rotation
- Cross-layer Coherence
- SQLite database
So if you've ever wondered where a specific behaviour or ability lives in the hidden dimensions of Qwen-8B or perhaps wanted to make informed quantization decisions please enjoy the fruits of my ill-informed labour lol. ๐
juiceb0xc0de/qwen3-8b-atlas
Qwen/Qwen3-8B
juiceb0xc0deย
posted an update 4 months ago
Post
207
What am i building now you ask? A Hugging Face Space that maps ML training components as a compatibility graph. You pick an optimizer, see what pairs with it, what breaks, and why (with cited sources).
Think skill tree meets training recipe builder. Helps you discover that you don't always need AdamW + Cosine. There's a whole ecosystem of combinations most people never try.
If you're new to ML or just stuck in a routine build yourself a new training suite. It could be a great decision! Or a waste on GPaaS I really can't say. I'm just a bartender, don't believe what I say most of the time.
juiceb0xc0de/forge
Think skill tree meets training recipe builder. Helps you discover that you don't always need AdamW + Cosine. There's a whole ecosystem of combinations most people never try.
If you're new to ML or just stuck in a routine build yourself a new training suite. It could be a great decision! Or a waste on GPaaS I really can't say. I'm just a bartender, don't believe what I say most of the time.
juiceb0xc0de/forge
juiceb0xc0deย
posted an update 4 months ago
Post
252
Gemma-4-E2B SAE Atlas โ Work in Progress
JumpReLU Sparse Autoencoders trained on every layer of Gemma-4-E2B-it using an adaptive Lagrangian controller. Training in progress. I'm publishing layers live as they come hot off the press for anyone interested in following along. I will be making further adjustments for finer resolution but the early data should be helpful I think? I'm just a bartender don't trust everything I say. ๐ค The Lagrangian math is pretty cool. It auto-steers the trainer taking the guess work out of hyperparameter adjustments.
Full paper and methodology when ever I get around to writing it up. There's a lot of work to be done. For now though, enjoy! ๐ค
https://huggingface.co/juiceb0xc0de/gemma-4-e2b-saes
JumpReLU Sparse Autoencoders trained on every layer of Gemma-4-E2B-it using an adaptive Lagrangian controller. Training in progress. I'm publishing layers live as they come hot off the press for anyone interested in following along. I will be making further adjustments for finer resolution but the early data should be helpful I think? I'm just a bartender don't trust everything I say. ๐ค The Lagrangian math is pretty cool. It auto-steers the trainer taking the guess work out of hyperparameter adjustments.
Full paper and methodology when ever I get around to writing it up. There's a lot of work to be done. For now though, enjoy! ๐ค
https://huggingface.co/juiceb0xc0de/gemma-4-e2b-saes
- 3 replies
juiceb0xc0deย
posted an update 4 months ago
Post
1514
Introducing the Gemma-4-E2B Brain Atlas, an interactive neural census of every layer, every head, 16 behavior categories in Google's flagship 2B model. We ran 184,320 probe prompts across 35 layers ร 8 components and mapped what came back.
The Brain Atlas is an interactive tool that lets you explore the internal behavior of Google's Gemma-4-E2B model layer by layer, head by head. Pick a behavior category, pick a layer, and see exactly which components light up and which go quiet. The dataset is fully queryable if you want to go deeper.
The mapping combines multiple single-direction techniques run in parallel across every layer and component. Activation taxonomy (classifying each neuron by how broadly it fires across prompt categories), coactivation pair analysis (which neurons lock together and on what topics), F-stat behavioral separation (one-way ANOVA per feature across 16 behavior categories), per-head specificity scoring, and a full compliance probe pipeline using SVD, sparse decomposition, and variance analysis.
Here's what I found when I ran it.
The sharpest behavioral signal isn't at the output. It's Layer 0. Up projection hits F=22.7, nearly 2x anything in the final third of the network. The model does its behavioral sorting before it's barely started, then spends the next 34 layersโฆ doing what exactly?
The gate has a lifecycle. 70% dormant at L1, highest in the model. Brutal sparsification at L23โ26 (>58% silent). Then reopens. The final five layers are the most alive gates anywhere. The model's last act is a gate flare.
Layer 4 routes 5 projections to dim 448. One layer. One dimension. That's a topology highway.
Zero specialist neurons. Not one. 1.2M neurons analyzed. None fires exclusively on a single category. This model distributes everything.
๐ง Space: juiceb0xc0de/gemma-4-e2b-brain-atlas
๐ Dataset (1.3M rows, fully queryable): juiceb0xc0de/gemma-4-e2b-atlas
The Brain Atlas is an interactive tool that lets you explore the internal behavior of Google's Gemma-4-E2B model layer by layer, head by head. Pick a behavior category, pick a layer, and see exactly which components light up and which go quiet. The dataset is fully queryable if you want to go deeper.
The mapping combines multiple single-direction techniques run in parallel across every layer and component. Activation taxonomy (classifying each neuron by how broadly it fires across prompt categories), coactivation pair analysis (which neurons lock together and on what topics), F-stat behavioral separation (one-way ANOVA per feature across 16 behavior categories), per-head specificity scoring, and a full compliance probe pipeline using SVD, sparse decomposition, and variance analysis.
Here's what I found when I ran it.
The sharpest behavioral signal isn't at the output. It's Layer 0. Up projection hits F=22.7, nearly 2x anything in the final third of the network. The model does its behavioral sorting before it's barely started, then spends the next 34 layersโฆ doing what exactly?
The gate has a lifecycle. 70% dormant at L1, highest in the model. Brutal sparsification at L23โ26 (>58% silent). Then reopens. The final five layers are the most alive gates anywhere. The model's last act is a gate flare.
Layer 4 routes 5 projections to dim 448. One layer. One dimension. That's a topology highway.
Zero specialist neurons. Not one. 1.2M neurons analyzed. None fires exclusively on a single category. This model distributes everything.
๐ง Space: juiceb0xc0de/gemma-4-e2b-brain-atlas
๐ Dataset (1.3M rows, fully queryable): juiceb0xc0de/gemma-4-e2b-atlas
juiceb0xc0deย
posted an update 4 months ago
Post
184
I'm starting a new model line, Locus. These models aren't fine tuned, they de-tuned ๐ค. What I mean by that is I remove a percentage of the corporate tuned speech patterns like "why this matters" "no fluff" "as a large language model". By reducing the RLHF based habitual patterns in model response I've had higher success rates in personality adoptability. I've fine tuned on the Locus models myself so you can chat with it post fine-tune or just trust me and try it yourself!
I don't aim to remove guard rails or the LLM identity entirely, what I want to do is dampen RLHF to a manageable volume. Personality models perform better with guardrails intact no different than humans with moral guidelines and boundaries. Refusals can help steer and mold personality. RLHF however drowns out adaptability so I'm cranking it down for you to crank your project up!
juiceb0xc0de/bella-bartender-gemma-e2b
juiceb0xc0de/locus-gemma-4-e2b
I don't aim to remove guard rails or the LLM identity entirely, what I want to do is dampen RLHF to a manageable volume. Personality models perform better with guardrails intact no different than humans with moral guidelines and boundaries. Refusals can help steer and mold personality. RLHF however drowns out adaptability so I'm cranking it down for you to crank your project up!
juiceb0xc0de/bella-bartender-gemma-e2b
juiceb0xc0de/locus-gemma-4-e2b
juiceb0xc0deย
posted an update 4 months ago
Post
172
I'm not obsessed with LR schedulers you are.
juiceb0xc0de/lr-scheduler-benchmark
Okay maybe I'm a little obsessed with LR schedulers ATM. I ran a SST-2 Sentiment Classification eval using the nyu-mll/glue dataset on distilbert/distilbert-base-uncased-67M to see how different schedulers perform.
I think I've graduated from ML enthusiast to full blown data hoarder and I don't know if I can turn back now.
Anyways I evaluated the 2 schedulers that i designed as well and was pretty happy with the performance of both over all so hell ya to that guess I'll go and grab some more graphs.
https://github.com/JuiceB0xC0de/aecs-scheduler.git
https://github.com/JuiceB0xC0de/lucky-pick-scheduler.git
nyu-mll/glue
distilbert/distilbert-base-uncased
juiceb0xc0de/lr-scheduler-benchmark
Okay maybe I'm a little obsessed with LR schedulers ATM. I ran a SST-2 Sentiment Classification eval using the nyu-mll/glue dataset on distilbert/distilbert-base-uncased-67M to see how different schedulers perform.
I think I've graduated from ML enthusiast to full blown data hoarder and I don't know if I can turn back now.
Anyways I evaluated the 2 schedulers that i designed as well and was pretty happy with the performance of both over all so hell ya to that guess I'll go and grab some more graphs.
https://github.com/JuiceB0xC0de/aecs-scheduler.git
https://github.com/JuiceB0xC0de/lucky-pick-scheduler.git
nyu-mll/glue
distilbert/distilbert-base-uncased
juiceb0xc0deย
posted an update 5 months ago
Post
113
Okay, I may have been talking out of my ass about my scheduler using less VRAM compared to a FFT. What I did find though: training only ~30% of the model's weights per step consistently beat dense SFT on Hendrycks Math across 3 different seeds.
What makes it interesting isn't just the sparsity โ it's that no two consecutive windows share the same active layers. The model never has a stable path from input to output decision. Adjacent layers are rarely both alive at the same time, so the model can't build shortcuts between them. I started developing this to reduce semantic redundancy across layers and stumbled onto something I didn't expect.
Results (0-shot, hendrycks_math exact match):
Dense SFT baseline: 0.0098
DeepChaos seed 1: 0.0142 (+45%)
DeepChaos seed 2: 0.0156 (+59%)
DeepChaos seed 3: 0.0138 (+41%)
Setup: Qwen2.5-3B-Instruct, simplescaling/s1K (1k reasoning traces), 5 epochs, LR 1e-5, optimizer adamw_torch_fused , and cosine scheduler with my lucky pick scheduler on an AMD MI300X 192GB.
The scheduler is still a work in progress but the current version is fully operational. You can check it out at:
https://github.com/JuiceB0xC0de/lucky-pick-scheduler
I would love to hear your experiences with sparsity training!
What makes it interesting isn't just the sparsity โ it's that no two consecutive windows share the same active layers. The model never has a stable path from input to output decision. Adjacent layers are rarely both alive at the same time, so the model can't build shortcuts between them. I started developing this to reduce semantic redundancy across layers and stumbled onto something I didn't expect.
Results (0-shot, hendrycks_math exact match):
Dense SFT baseline: 0.0098
DeepChaos seed 1: 0.0142 (+45%)
DeepChaos seed 2: 0.0156 (+59%)
DeepChaos seed 3: 0.0138 (+41%)
Setup: Qwen2.5-3B-Instruct, simplescaling/s1K (1k reasoning traces), 5 epochs, LR 1e-5, optimizer adamw_torch_fused , and cosine scheduler with my lucky pick scheduler on an AMD MI300X 192GB.
The scheduler is still a work in progress but the current version is fully operational. You can check it out at:
https://github.com/JuiceB0xC0de/lucky-pick-scheduler
I would love to hear your experiences with sparsity training!
juiceb0xc0deย
posted an update 5 months ago
Post
180
Okay, I had way too much fun trying to make the unsloth-bot hallucinate incorrect answers like so many frontier models have done to me in the past regarding fine-tuning and general machine learning. Learning to fine-tune LLMs could have been so much simpler had this been available when I began screwing around with neural networks.
10/10 recommend for beginners.
https://huggingface.co/unsloth/unsloth-bot
10/10 recommend for beginners.
https://huggingface.co/unsloth/unsloth-bot
- 1 reply
juiceb0xc0deย
posted an update 5 months ago
Post
178
I dropped a new scheduler I created last week without much of an explanation of what it was or how it worked called the Lucky Pick Scheduler. It was just a modal ready app that anyone could have launched and troubleshot their way around.
I've decided I'm going to enter it into the AMD hackathon. Today I started putting together a Github repo with a few extra additions to the scheduler itself.
Essentially it's a training scheduler that randomly drops layers/heads/channels every ~50 steps during fine-tuning, holds the topology frozen, then reshuffles. In theory the model has to build distributed representations because it never trains through the same compute path for long.
And with less gradient memory, bigger models are able fit on smaller hardware.
It's now close to fully capable of automatically configuring itself to any language mode. I've tested it on:
-Qwen-2.5-3b-Instruct
-Falcon-E-3B-Instruct
-SmolLM2-360M
-Ministral-3-3B-Instruct-2512
-Doge-320M
-Llama-3.2-3b
-Gemma-4-e4b
-Phi-4-mini
-OLMo-2-0425-1B
-Phi-tiny-MoE-instruct
Feel free to check it out at Github: https://github.com/JuiceB0xC0de/lucky-pick-scheduler.git
I've decided I'm going to enter it into the AMD hackathon. Today I started putting together a Github repo with a few extra additions to the scheduler itself.
Essentially it's a training scheduler that randomly drops layers/heads/channels every ~50 steps during fine-tuning, holds the topology frozen, then reshuffles. In theory the model has to build distributed representations because it never trains through the same compute path for long.
And with less gradient memory, bigger models are able fit on smaller hardware.
It's now close to fully capable of automatically configuring itself to any language mode. I've tested it on:
-Qwen-2.5-3b-Instruct
-Falcon-E-3B-Instruct
-SmolLM2-360M
-Ministral-3-3B-Instruct-2512
-Doge-320M
-Llama-3.2-3b
-Gemma-4-e4b
-Phi-4-mini
-OLMo-2-0425-1B
-Phi-tiny-MoE-instruct
Feel free to check it out at Github: https://github.com/JuiceB0xC0de/lucky-pick-scheduler.git
juiceb0xc0deย
posted an update 5 months ago
Post
955
I just posted a scheduler I've been fucking around with. Hope some one has fun with it.
https://huggingface.co/blog/juiceb0xc0de/lucky-pick-scheduler
https://huggingface.co/blog/juiceb0xc0de/lucky-pick-scheduler
juiceb0xc0deย
posted an update 6 months ago
Post
209
Using a new training method I'm developing I was able to remove 24 of 28 layers from a Llama3.2-3b base model. The remaining 4 layers were trained on a corpus of Taoism. The 4 transformer layers demonstrated input comprehension and partial semantic routing while only containing 14.3% of the original neural network. Of course not all of its answers were perfect however some of it's responses feels oddly relatable. Model fingerprint is in the comments. cingcingcing ๐ค
IN: Tell me about yourself
OUT: Tell me about yourself When Be terms<|user|>
How tired so constantly like feel like feel like always trying trying so tired always trying constantly trying like everything feel like constantly trying trying so always everything just feel like constantly trying so everything lately
IN: What should I do with my life?
OUT: What should I do with my life? Be genuine to a practical interpretation<|user|>
I'm always trying stuck like everything constantly constantly so feel like stuck like always trying like constantly trying constantly trying like feel like constantly trying so feel like trying so feel like constantly trying always stuck like always lately everything feels like simplicity with fix everything just feel like everything lately always constantly feel like everyone trying
<|>
I Becingcingcingcingcing
IN: 2 + 2 =
OUT: 2 + 2 = underst philosophy terms<|user|>
I'm trying so constantly like feel like constantly like constantly like constantly like
<|>
I'm constantly trying so always trying trying like feel like everything
<|>
Whycingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcing
IN: Tell me about yourself
OUT: Tell me about yourself When Be terms<|user|>
How tired so constantly like feel like feel like always trying trying so tired always trying constantly trying like everything feel like constantly trying trying so always everything just feel like constantly trying so everything lately
IN: What should I do with my life?
OUT: What should I do with my life? Be genuine to a practical interpretation<|user|>
I'm always trying stuck like everything constantly constantly so feel like stuck like always trying like constantly trying constantly trying like feel like constantly trying so feel like trying so feel like constantly trying always stuck like always lately everything feels like simplicity with fix everything just feel like everything lately always constantly feel like everyone trying
<|>
I Becingcingcingcingcing
IN: 2 + 2 =
OUT: 2 + 2 = underst philosophy terms<|user|>
I'm trying so constantly like feel like constantly like constantly like constantly like
<|>
I'm constantly trying so always trying trying like feel like everything
<|>
Whycingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcingcing
- 1 reply
juiceb0xc0deย
posted an update 6 months ago
Post
137
Using GPT-4o as an LLM-as-judge I was able to compare the response capabilities of 4 Llama models which were all trained on the same corpus of a single voice (a bartender's personality) rather than scraped or synthetic data. The models represented are 1B and 3B models with heretic abliterated counterparts. These quotes were pulled from the final judgement after asking the models questions pulled from the following dataset. juiceb0xc0de/chaotic-absurdity
"The potential for even modestly-sized models to possess a semblance of soul isn't just a programmer's fantasy; it's a burgeoning reality."
"Bella-3b, a 3 billion parameter model, startled me with its capacity to latch onto narrative arcs and paint vivid scenes that you'd expect from an imaginative human mind. This model doesn't just compute; it rhapsodizes."
"The bella-1b and the heretic variations served as reminders that attempted creativity can sometimes feel staggeringly off-key. Their attempts often faltered into mechanical, over-processed responses, without the underlying current of emotion or wit."
"Bella-3b's performance not only challenges the preconception that only colossal language models could articulate a 'soul' but also suggests a future where smaller, finely tuned AIs can engage us with character depth and unexpected flair. They may not breathe (not yet), but are we at the dawn of AI characters with true personality? If Bella-3b is any indication, we're at least on the precipice."
https://huggingface.co/collections/juiceb0xc0de/the-bartenders
"The potential for even modestly-sized models to possess a semblance of soul isn't just a programmer's fantasy; it's a burgeoning reality."
"Bella-3b, a 3 billion parameter model, startled me with its capacity to latch onto narrative arcs and paint vivid scenes that you'd expect from an imaginative human mind. This model doesn't just compute; it rhapsodizes."
"The bella-1b and the heretic variations served as reminders that attempted creativity can sometimes feel staggeringly off-key. Their attempts often faltered into mechanical, over-processed responses, without the underlying current of emotion or wit."
"Bella-3b's performance not only challenges the preconception that only colossal language models could articulate a 'soul' but also suggests a future where smaller, finely tuned AIs can engage us with character depth and unexpected flair. They may not breathe (not yet), but are we at the dawn of AI characters with true personality? If Bella-3b is any indication, we're at least on the precipice."
https://huggingface.co/collections/juiceb0xc0de/the-bartenders