PraxaLing
๐ฆ
2
Practice a language with interactive AI conversations
Reuben fernandes PRO
Reubencf
AI & ML interests
LLM
Recent Activity
updated a Space 11 minutes ago
Reubencf/PraxaLing reacted to theirpost with ๐ฅ about 13 hours ago
https://huggingface.co/spaces/Reubencf/HF_QR_Gen
has been improved create custom Qr-codes posted an update about 13 hours ago
https://huggingface.co/spaces/Reubencf/HF_QR_Gen
has been improved create custom Qr-codesOrganizations
updated a
Space 11 minutes ago
reacted to their post with ๐ฅ about 13 hours ago
updated a
Space about 14 hours ago
reacted to RDTvlokip's post with ๐ 3 days ago
Post
1979
I finally changed the architecture of my 15M French LLM. It worked. Then I almost fooled myself about how much and catching that was the real win.
After proving last time that architecture is a threshold, not a lever, I got stubborn: could I change how the model learns? Four honest attempts, Lion, a sharper AdamW ฮฒ2, multi-token prediction, LayerScale. Four failures. The bottleneck wasn't the learning rule either.
So I changed the shape of the computation instead: loop the same transformer blocks 4ร, deeper reasoning, zero added parameters. It beat the baseline on perplexity, the first thing in the whole project to move that number. Then I added my own twist: let each token decide how deep to think, halting on its own entropy.
My first evaluation was spectacular. Coherence up 65%. Hallucinated names down 62%.
It was noise.
Eight prompts, one seed. I re-ran on 50 prompts ร 200 tokens and watched the gains shrink to "modest" and on out-of-domain prompts, recurrence actually made things worse. No universal winner. And none of it is new: it's Adaptive Computation Time (2016), the Universal Transformer (2018), and LoopViT (2026), recombined and measured honestly.
The real lesson:
A number from 8 prompts is a rumor. The eval harness that kills your own best result is worth more than the result it kills. Cite your lineage. Stay preliminary until multiple seeds say otherwise.
The three models are live. The write-up is honest about every caveat ๐
๐ https://huggingface.co/blog/RDTvlokip/teaching-a-15m-french-llm-to-think-deeper
After proving last time that architecture is a threshold, not a lever, I got stubborn: could I change how the model learns? Four honest attempts, Lion, a sharper AdamW ฮฒ2, multi-token prediction, LayerScale. Four failures. The bottleneck wasn't the learning rule either.
So I changed the shape of the computation instead: loop the same transformer blocks 4ร, deeper reasoning, zero added parameters. It beat the baseline on perplexity, the first thing in the whole project to move that number. Then I added my own twist: let each token decide how deep to think, halting on its own entropy.
My first evaluation was spectacular. Coherence up 65%. Hallucinated names down 62%.
It was noise.
Eight prompts, one seed. I re-ran on 50 prompts ร 200 tokens and watched the gains shrink to "modest" and on out-of-domain prompts, recurrence actually made things worse. No universal winner. And none of it is new: it's Adaptive Computation Time (2016), the Universal Transformer (2018), and LoopViT (2026), recombined and measured honestly.
The real lesson:
A number from 8 prompts is a rumor. The eval harness that kills your own best result is worth more than the result it kills. Cite your lineage. Stay preliminary until multiple seeds say otherwise.
The three models are live. The write-up is honest about every caveat ๐
๐ https://huggingface.co/blog/RDTvlokip/teaching-a-15m-french-llm-to-think-deeper
- 6 replies
reacted to their post with ๐ง ๐คฏ๐ 3 days ago
Post
134
Introducing Reubencf/Document_Query
powered by CohereLabs/command-a-plus-05-2026-w4a4 by cohere
just open admin page -> drop in any document -> and ask questions
powered by CohereLabs/command-a-plus-05-2026-w4a4 by cohere
just open admin page -> drop in any document -> and ask questions
- 1 reply
reacted to pankajpandey-dev's post with โค๏ธ๐๐ฅ 3 days ago
Post
4042
๐ฎ๐ณ Qwen3.5-9B Hindi Instruct โ it stops thinking in English
Ask base Qwen3.5-9B a question in Hindi and it burns hundreds of tokens thinking in English inside its think block before a single Devanagari word appears โ then code-switches in the answer. I fine-tuned it to close the think block instantly and reply in pure, native Hindi.
โ Model (16-bit): pankajpandey-dev/qwen3.5-9b-hindi-instruct
โ GGUF (Q4/Q5/Q8): pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF
โ Try it in the browser: pankajpandey-dev/qwen3.5-9b-hindi-demo
Recipe: Unsloth + LoRA (r=16, response-only loss) on 12.9k Hindi pairs โ AI4Bharat anudesh + dolly-hi + wikiHow-hi + Aya Hindi (human-written). The Q4_K_M is 5.4 GB and runs on a plain laptop CPU.
New in this run vs my earlier models: mixed in long-form native sources (wikiHow) after my last eval showed the fine-tune traded detail for conciseness โ this one keeps answers detailed and native.
Part of my weekly ๐ฎ๐ณ Hindi LLM Series. Feedback welcome ๐
#Hindi #IndicNLP #Qwen #GGUF #LocalLLM #Unsloth
Ask base Qwen3.5-9B a question in Hindi and it burns hundreds of tokens thinking in English inside its think block before a single Devanagari word appears โ then code-switches in the answer. I fine-tuned it to close the think block instantly and reply in pure, native Hindi.
โ Model (16-bit): pankajpandey-dev/qwen3.5-9b-hindi-instruct
โ GGUF (Q4/Q5/Q8): pankajpandey-dev/qwen3.5-9b-hindi-instruct-GGUF
โ Try it in the browser: pankajpandey-dev/qwen3.5-9b-hindi-demo
Recipe: Unsloth + LoRA (r=16, response-only loss) on 12.9k Hindi pairs โ AI4Bharat anudesh + dolly-hi + wikiHow-hi + Aya Hindi (human-written). The Q4_K_M is 5.4 GB and runs on a plain laptop CPU.
New in this run vs my earlier models: mixed in long-form native sources (wikiHow) after my last eval showed the fine-tune traded detail for conciseness โ this one keeps answers detailed and native.
Part of my weekly ๐ฎ๐ณ Hindi LLM Series. Feedback welcome ๐
#Hindi #IndicNLP #Qwen #GGUF #LocalLLM #Unsloth
- 4 replies
replied to their post 4 days ago
@CohereLabs
upvoted a collection 5 days ago
updated a
collection 5 days ago
updated a
Space 5 days ago
updated a
model 5 days ago
published a
model 5 days ago
published a
Space 5 days ago